1.91.31,

#3:“: t ‘ . . . . ‘ , .
. . . ‘ . ‘ , ‘ . . ‘ :1
, ‘ . . . . 9‘ Law.
. . . a3" ..
. . .A «x... ..

\IJiQUW

. a...
#1253
k’ll‘l (5- In"

.. . :...

xr .
an 3.64

a.

 

. .Apx...‘ C... —. :va‘.1.
. . 2

 

mark-aux §¥i§$v ., -18.) .. , V , ‘ ,
. . . .i. .. E .&.u4u.rufl., . . . . x , 9. ‘
J . ‘ ”a: ”.3”? “mug

1.44.4 I.
c P.

 

(9 LIBRARY
ass-7 Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

Assessing Physical Activity Behaviors in College Students

presented by

Joshua James Ode

has been accepted towards fulfillment
of the requirements for the

 

 

PhD. degree in Kinesiology

 

 

a Major Professor’s Signature

3/7, 07

 

Date

MSU is an afiVnnetive-action, equal-opportunity employer

 

--.-.-.—.-.--—.—._._._V-.-A._ 4

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:lClRC/DateDue.indd-p.1

ASSESSING PHYSICAL ACTIVITY BEHAVIORS IN COLLEGE STUDENTS
By

Joshua James Ode

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Kinesiology

2007

ABSTRACT
ASSESSING PHYSICAL ACTIVITY BEHAVIORS IN COLLEGE STUDENTS
By

Joshua James Ode

Physical activity declines during the transition from adolescence to adulthood. In
order to evaluate this transition, it is important to assess physical activity behaviors in the
young adult population. This is possible by assessing college students, which represent a
large proportion of the young adult population. Currently, the determinants of physical
activity participation in college students are poorly understood and a more detailed
understanding is needed to help prevent the decline in physical activity. PURPOSE: The
purpose of this study was to assess the impact of gender, enrollment in a college activity-
based class, and high school physical activity participation on physical activity behaviors
in college students. A second purpose to this study was to assess the impact of
enrollment in a healthy lifestyles course on change in physical activity during a semester.
METHODS: A total of 911 college students enrolled in a healthy lifestyles class (n=455)
and communications class (n=456) completed the baseline survey. A total of 765
students (healthy lifestyles = 355, communications = 365) completed the follow-up
survey at the end of the semester. Study participants completed an intemet questionnaire
assessing frequency, intensity, and duration of physical activity. Physical activity energy
expenditure was specified as caloric expenditure indexed by body weight (kcal/kg/week)
and quartiles of kcal/kg/week were used for analyses (Quartile 1: 0-125 kcal/kg/week,
Quartile 2: 12.5-25.5 kcal/kg/week, Quartile 3: 25.5-45.9 kcal/kg/week, Quartile 4: >459

kcal/kg/week). A change score was used to assess the differences in physical activity at

the beginning and end of the semester (Decrease: move to a lower quartile, Stay the
Same: stay in the same quartile, Increase: increase to a higher quartile during the
semester). Cross-tabulations were generated and chi square was used to determine any
significant associations (p<.05) between physical activity participation and each exposure
variable. Multinomial logistic regression was used to determine the magnitude of these
associations. RESULTS: When compared to the lowest physical activity quartile, males
had 7.2 higher odds than females of belonging to the highest physical activity quartile.
There was no difference in physical activity between students enrolled in the healthy
lifestyles class and communications class. College students who played 1 high school
sport, 2 sports, 3 sports, or 4 or more sports were 2.12, 3.3, 3.1, and 5.2 times more likely
to be in the highest physical activity quartile. Students who were classified as moderately
active during high school leisure time physical activities were 2.5 times more likely to be
in quartile 2, 3.7 times more likely to be in quartile 3, and 9.9 times more likely to be in
quartile 4. Students enrolled in the healthy lifestyles class had 81% higher odds than
communication students to increase physical activity. CONCLUSIONS: Despite the
concern regarding physical inactivity on college campuses, little is known about physical
activity in this population. Therefore, the results of this study provide valuable
information about the determinants of physical activity in the college students and that
enrollment in a healthy lifestyles class may have a positive impact on physical activity

participation.

Copyright by
JOSHUA JAMES ODE
2007

ACKNOWLEDGEMENTS

I would like to first thank my advisor, Jim Pivanik for his direction and
mentorship throughout the past four years. Your dedication to your graduate students is
unparalleled. I would also like to thank Dr. Mat Reeves for your time, effort, and for
continually challenging me to think critically about research, Dr. Jim Anthony for the
research opportunities you have provided, and Dr. Dan Gould for teaching me the
importance of sport psychology. I would like to thank Jeremy Knous and Lanay Mudd
for your contributions and support throughout this project and JoAnn Janes for always
having the answer. Finally, I would like to thank my parents for their unconditional
support, my wife Christina for your encouragement, understanding, and love, and my son

Tyler for keeping me grounded.

Acknowledgment of funding support:
0 This research was supported by a Student Awards Program grant fiom the Blue
Cross and Blue Shield of Michigan Foundation.
0 This research was supported by a Plus One Active Research Grant on Wellness

Using the Internet from the American College of Sports Medicine Foundation.

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... viii
LIST OF FIGURES ............................................................................................................ x
CHAPTER 1 INTRODUCTION ......................................................................................... 1
CHAPTER 2 REVIEW OF LITERATURE ........................................................................ 9
2.1 Introduction ........................................................................................................... 9
2.2 Physical Activity ................................................................................................... 9
2.3 Physical Activity Assessment ............................................................................. 11
Introduction ...................................................................................................... l 1
Physical Activity Questionnaire ...................................................................... ll
Reliability and Validity of Physical Activity Questionnaires .......................... 13
Conclusion ....................................................................................................... 19
2.4 Physical Activity in College Students ................................................................. 19
Introduction ...................................................................................................... 19
Review Studies of College Students’ Physical Activity Levels ...................... 20
Physical Activity Participation in College Students ........................................ 22
Physical Activity Levels among College Students Enrolled in a
Health Related Class ........................................................................................ 31
2.5 Decline in Physical Activity from High School into College ............................. 36
Introduction ...................................................................................................... 36
Physical Activity during the Transition from High School to College ............ 37
Changes in Physical Activity between Adolescence and
Young Adulthood ............................................................................................. 40
Conclusion ....................................................................................................... 45
2.6 Changes in Physical Activity During College .................................................... 45
CHAPTER 3 METHODS .................................................................................................. 48
3.1 Research Design .................................................................................................. 48
3.2 Study Population ................................................................................................. 49
3.3 Eligibility Criteria (inclusion/exclusion) ............................................................ 50
3.4 Study Participants ............................................................................................... 50
3 .5 Recruitment Procedures ...................................................................................... 50
3.6 Over-Arching Conceptual Model ....................................................................... 52
3.7 Assessment Protocol ........................................................................................... 54
3.8 Primary Outcome Variables ................................................................................ 56
3.9 Exposure Variables ............................................................................................. 58
3.10 Study Aim 1 ...................................................................................................... 62
3.11 Statistical Analysis for Study Aim 1 ................................................................. 62
3.12 Study Aim 2 ...................................................................................................... 66
3.13 Statistical Analysis for Study Aim 2 ................................................................. 66

vi

3.14 Supplementary Statistical Analysis ................................................................... 68

CHAPTER 4 RESULTS .................................................................................................... 71
4.1 Descriptive Data .................................................................................................. 71
4.2 Description of Study Aim 1 ................................................................................ 74
4.3 Assessment of Physical Activity Outcome Variable .......................................... 75
4.4 Study Aim 1 Descriptive Results ........................................................................ 76
4.5 Test of the Proportional Odds Assumption ......................................................... 79
4.6 Study Aim 1 Bivariate Ordinal Logistic Regression Results .............................. 81
4.7 Study Aim 1 Bivariate Multinomial Logistic Regression Results ...................... 84
4.8 Study Aim 1 Adjusted Multinomial Logistic Regression Results ...................... 88
4.9 Description of Study Aim 2 ................................................................................ 91
4.10 Time Varying Covariates .................................................................................. 92
4.11 Study Aim 2 Descriptive Results ...................................................................... 92
4.12 Unadjusted Bivariate Results ............................................................................ 95
4.13 Study Aim 2 Adjusted Multinomial Logistic Regression Results .................... 96
4.14 Results for Supplementary Statistical Analyses ............................................... 98
CHAPTER 5 DISCUSSION ............................................................................................ 103
5.1 Effect of Gender and Class on Physical Activity Participation ........................ 103
5.2 Effect of High School Sports and Other Leisure Time Physical
Activities on Physical Activity in College .............................................................. 107
5.3 Effect of Enrollment in a Healthy Lifestyles Class on Change in
Physical Activity ..................................................................................................... 110
5.4 Assessment of the Effect of Using Different Specifications of
Physical Activity Energy Expenditure .................................................................... 113
5.5 Assessment of Supplemental Analysis Results on the Impact of
Enrollment in a Healthy Lifestyles Class ................................................................ 117
5.6 Future Studies of Physical Activity in College Students .................................. 120
5.7 Conclusion ........................................................................................................ 122
APPENDIX A Consent Form .......................................................................................... 123
APPENDIX B Questionnaire ........................................................................................... 124
APPENDIX C Log Transformation Data for Kcal/kg/week ........................................... 129
APPENDIX D Results for Continuous Physical Activity Outcome Variable ................. 131
APPENDIX E Results for Categorical Physical Activity Outcome Variable ................. 133
APPENDIX F Study Aim 3 Research Design ................................................................. 136
REFERENCES ................................................................................................................ 147

vii

LIST OF TABLES

Table 1. Descriptive data comparison for the healthy lifestyles classes ............................ 72
Table 2. Descriptive data for the study population ............................................................ 74

Table 3. Descriptive data for quartiles of physical activity at the beginning of a
semester .............................................................................................................................. 78

Table 4. Unadjusted ordinal logistic regression results for quartiles of physical
activity at the beginning of a semester ............................................................................... 83

Table 5. Unadjusted multinomial logistic regression results ............................................. 87

Table 6. Adjusted multinomial logistic regression results for quartiles of physical
activity at the beginning of a semester based on high school physical activity ................. 90

Table 7. Descriptive data for change in physical activity quartile ..................................... 94

Table 8. Unadjusted multinomial logistic regression results for change in quartile of
physical activity during a semester .................................................................................... 96

Table 9. Adjusted multinomial logistic regression results for change in quartile of
physical activity during a semester .................................................................................... 98

Table 10. Estimated relationship between physical activity energy expenditure
and class membership during enrollment in a healthy lifestyles class
(population-based approach) .............................................................................................. 99

Table 11. Estimated relationship between physical activity energy expenditure
and class membership during enrollment in a healthy lifestyles class
(subject-specific approach) ................................................................................................ 99

Table 12. Estimated relationship between physical activity energy expenditure
at follow-up and class membership afier controlling for physical activity energy
expenditure at baseline ..................................................................................................... 100

Table 13. Estimated relationship between physical activity energy expenditure
at follow-up and class membership after controlling for physical activity energy
expenditure at baseline and the nested structure of the healthy lifestyles class ............... 101

Table 14. Estimated relationship between physical activity energy expenditure and
enrollment in the fall or spring healthy lifestyles classes ................................................ 102

viii

Table 15. Mean, skewness, and kurtosis data for the continuous physical activity
outcome variable .............................................................................................................. 129

Table 16. Mean, skewness, and kurtosis data for the log transformed

continuous physical activity outcome variable ................................................................ 130
Table 17. Descriptive data for the log kcal/kg/wk .......................................................... 131
Table 18. Multiple regression analysis results for the log kcal/kg/wk ............................ 132
Table 19: Descriptive data for meeting the recommendations for physical activity ...... 133

Table 20. Unadjusted odds ratios for meeting the recommendations for
physical activity ............................................................................................................... 134

Table 21. Adjusted odds ratios for meeting the recommendations for
physical activity ............................................................................................................... 135

ix

LIST OF FIGURES

Figure 1. Timeline for data collection ................................................................................ 49
Figure 2. Over-arching conceptual model for Study Aim 1 .............................................. 53
Figure 3. Over-arching conceptual model for Study Aim 2 .............................................. 54
Figure 4. Graphical representation of physical activity quartile data ............................... 76

Figure 5. Graphical representation of the proportional odds assumption for the
unadjusted analysis of race (African American vs. Caucasian) ......................................... 80

Figure 6. Graphical representation of the proportional odds assumption for the
unadjusted analysis of class membership (Healthy Lifestyles vs. Communications) ....... 80

Figure 7. Change in odds ratios for gender after adjustment for exposure variables
included in the adjusted multinomial logistic regression model ........................................ 91

Figure 8. Frequency distribution for the continuous physical activity
outcome variable .............................................................................................................. 129

Figure 9. Frequency distribution for the log transformed continuous
physical activity outcome variable ................................................................................... 130

Figure 10: Timeline for data collection for study participants entering
the cohort in September ................................................................................................... 137

Figure 11: Timeline for data collection for study participants entering
the cohort in January ........................................................................................................ 137

CHAPTER 1
INTRODUCTION

The benefits of physical activity on overall health and chronic disease prevention
are well known, yet many Americans are not physically active on a regular basis. Recent
surveillance data have shown that only 45% of US adults 72 and 69% of high school
students 7' met the American College of Sports Medicine’s and the Centers for Disease
Control and Prevention’s minimum recommendations for physical activity. The
recommendations are defined as either moderate activity (participating in activities that
did not make the individual sweat or breathe hard for 330 minutes on 35 of the past 7
days) or vigorous activity (participating in activities that made the individual sweat and
breathe hard for 320 minutes on 23 of the past 7 days). Furthermore, differences in
physical activity prevalence between high school students and adults suggest a decline in
physical activity as adolescents move into adulthood. This age related decline in physical
activity is well documented, but poorly understood '5’ 60’ 68. Therefore, a more detailed
understanding of the transition in physical activity behaviors from adolescence into
adulthood is needed.

College age students are a convenient population to study when evaluating the
transition between adolescents and adults, as they may be sufficiently representative of
the young adult population. According to the US Department of Education, 16.6 million
students were enrolled in colleges or universities in 2004 and 29% of all young adults
between ages 25-29 have completed a bachelors degree or higher '8. In addition, the
most recent US Census Bureau report showed that nearly 53% of the adult population has

been enrolled in some college based education 66. Due to the decline in physical activity

that occurs during young adulthood and the large number of young adults enrolled in
college, it is important to assess physical activity behaviors in college students.

The American College Health Association, which published the Healthy Campus
2010 report, has recognized the importance of assessing physical activity on college
campuses. The purpose of the Healthy Campus 2010 document is to establish national
college health objectives which serve as a basis for developing plans to improve college
student health. Healthy Campus 201 0 identified the Leading Health Indicators, which
reflect major health concerns among college students and included physical inactivity as
a priority health risk behavior 5 . To further illustrate the importance of studying physical
activity behaviors in college students, results from a survey of recent college alumni
found that nearly half of all college students reported being less active six years following
graduation 65 . However, of the 44% of college students who reported exercising
regularly while attending college, 85% engaged in similar or higher levels of physical
activity following graduation 65. Therefore, there is a need for a detailed understanding
of college students’ physical activity behaviors, as an active lifestyle during college may
lead to an active lifestyle later in life.

Most physical activity and health related data collected in college students are

- I. .2 .4 .
cross-sectlonal 6" .12 8 3 58

and have yielded inconsistent results. For example,
prevalence estimates of college students who are physically active have ranged fi'om 33%
to 59.7% in previous studies. This wide range may be due in part to the methods used to
assess of physical activity, which often do not include questions needed to assess

intensity, duration, and frequency of exercise. In addition, small convenience samples

have been used which may not reflect the general college population. These criticisms

were identified in a recent meta-analysis addressing college student physical activity.

The authors concluded that comparing college students’ physical activity across studies is
difficult, if not impossible, and there is a need for additional research that includes data
on intensity, duration, and frequency of physical activity 36.

Despite the many limitations in previous studies, the prevalence of meeting the
recommendations for moderate and/or vigorous physical activity by college students is
consistently lower than in high school students. Of four previous studies that assessed
vigorous physical activity in college students, ” 44 26 48 the prevalence of college students
meeting the vigorous physical activity guidelines ranged from 42-53%. These results
were all substantially lower than the 65% prevalence rate of vigorous physical activity
reported in high school students participating in the Youth Risk Behavior Surveillance
System (YRBSS) which was developed to monitor the priority health risk behaviors
among nearly 11,000 youth in the United States. Previous surveillance data suggests the
prevalence of meeting the recommendations for moderate physical activity in college
students was only 21%, 48 which is 5% lower than high school students participating in
the YRBSS. Finally, a previous study reported that 58% of college students met the
recommendations for either moderate or vigorous activity 57 which is lower than the 69%
reported in the YRBSS. Interestingly, the prevalence rates for college students are
consistently higher than those found for US. adults participating in the Behavior Risk
Factor Surveillance System (BRF SS) which show a 25% prevalence for vigorous
physical activity and a 46% prevalence for vigorous or moderate physical activity. These

results illustrate the importance of assessing physical activity in college students as this

time period may reflect an important period stage in the age-related decline in physical
activity.

A potential way to improve the physical activity levels in college students and
potentially delay or reduce the age-related decline in physical activity is through
enrollment in physical activity and health related classes. Pearman et al. 53 performed a
cross-sectional study of 979 alumni fi'om two universities in which one required a
physical activity class and one did not. Although the cross-sectional design of this study
was a major limitation, alumni who attended the university where a physical activity
course was required weighed significantly less, were more likely to know their blood
pressure and cholesterol levels, and were more likely to engage in aerobic exercise than
those who attended the university where a physical activity course was not required.
These results suggest that enrollment in physical activity classes during college is
beneficial 53. Despite the potential benefits, few colleges or universities currently require
physical activity/healthy lifestyles classes for their students. In the 1960’s, nearly 90% of
college institutions required physical education courses. This number has decreased to
nearly 65% in the 1990’s 3 1. Often, a variety of elective physical activity and health
related courses are offered to college students. However, due to a limited number of
studies, data on the impact of these classes on physical activity levels are lacking 34. As a
result, there is a need to examine the effect of enrollment in physical activity classes on
college students’ physical activity behaviors. This information could be helpful in
improving the content of and/or promote enrollment in health-related physical activity

COlll'SCS.

An important transition period in the age related decline in physical activity may
occur as adolescents transition from high school to college. In order to illustrate specific
details of the decline in physical activity during this period, Bray and Born asked first
year college students about their current and past physical activity levels ”. Results
indicated that only 33% of respondents were active in both high school and college, while
another 33% were active during high school and became inactive during college 1'.
There are minimal data to support the physical activity decline from high school to
college as most studies are cross-sectional and were not designed to determine change in
physical activity over time. The majority of longitudinal research focusing on this
transition period assesses physical activity participation throughout adolescence and
young adulthood. Telema and Yang 68 found that between the ages of 12 and 27,
frequency of physical activity declined 57% and 28% in a large sample of males and
females, respectively. In addition, total moderate or vigorous physical activity (defined
by frequency and intensity) decreased 55% in males and 20% in females. Van Mechelen
and colleagues 73 reported similar results in Dutch males and females in which there was
a 31% (men) and 8% (women) decline in total time spent in habitual physical activity
between the ages of 13 and 27. Furthermore, decreases in weekly energy expenditure of
42% in males and 17% in females occurred over the same time period. Although this
research showed a decline in physical activity from adolescence to young adulthood,
there is a lack of research that specifically assesses the transition from high school to
university. This transition period is important as it is a unique experience for each
individual that may include academic, social, cultural, emotional, as well as physical

changes 24.

Understanding the determinants for the age-related decline in physical activity is
difficult as they may be specific to certain periods of life including the time spent at
college. To the best of our knowledge, no longitudinal studies have assessed changes in
physical activity and the corresponding determinants of these changes during the four
years of enrollment in college and following graduation. Thus, we have limited
understanding of the influence of college enrollment on physical activity levels, and there
is a need to develop research protocols aimed at assessing this relationship. One purpose
of this project is to develop a longitudinal research protocol which will effectively track
physical activity behavior and its determinants over four years of college and following
graduation.

Study Summary: This investigation involves a detailed assessment of college
student physical activity behaviors. Approximately 900 students were studied while they
were enrolled in a healthy lifestyles course or communications course at Michigan State
University. This dissertation focuses on a detailed assessment of college students’
physical activity levels. Currently, methods used to collect physical activity data in
college students are inconsistent and often lack detail. Study strengths include: 1) a
detailed assessment the frequency, intensity, duration, and type of physical activity, 2)
assessment of the relationship between physical activity and a variety of demographic
variables and other health related behaviors, 3) a retrospective assessment of high school
sport participation and physical activity behaviors and 4) a longitudinal assessment of
physical activity during enrollment in either a combined lecture/activity healthy lifestyles
class or a communications course. An underlying objective of this dissertation is to

determine if results change when using different physical activity outcome variables.

Study Aim 1 includes two Specific sub-aims. Sub-aim 1 focuses on the impact of
gender and membership in a healthy lifestyles class on physical activity behaviors at the
beginning of a semester. Sub-aim 2 focuses on the impact of high school sport and
physical activity participation on physical activity levels during college. The two sub-
aims for Study Aim 1 are as follows:

Study Aim 1 (sub-aim 1): Does gender or class membership influence physical
activity energy expenditure in college students?

Hypothesis: Male college students will be more physically active than female
college students. Students enrolled in the healthy lifestyles class will be more physically
active than students enrolled in the communications class.

Study Aim 1 (sub-aim 2): Does high school sport and physical activity
participation influence physical activity energy expenditure in college students?

Hypothesis: College students who participated in high school sports will be more
physically active than those who did not participate in high school sports. College
students who were more physically active outside of high school sports will be more
active than those who were not active outside of school sports.

The second aim of this dissertation focuses on the impact of enrollment in a four-
month combined lecture/activity healthy lifestyles class on change in physical activity
levels following completion of the class. The aim includes a comparison of the change in
physical activity between students enrolled in either the healthy lifestyles class or a
communications class: Study Aim 2 is as follows:

Study Aim 2: Does enrollment in a semester long healthy lifestyles class

influence changes in physical activity energy expenditure?

Hypothesis: At the end of the semester, college students enrolled in the healthy
lifestyles class will have greater physical activity energy expenditure compared to
students enrolled in the communications class.

The third aim of this dissertation focuses on the development of methodology that
will be used to collect longitudinal data on college students over a 4-year period. Data
will not be collected as part of the study design development.

Study Aim 3: To develop methodology needed to assess changes in physical
activity and health related behaviors during enrolhnent as an undergraduate college

student at a large Midwestern University.

CHAPTER 2
REVIEW OF LITERATURE

2.1 Introduction

The age related decline in physical activity across all ages is well documented, but
the reasons for this decline are poorly understood 60. Specifically, there appears to be a
decline in physical activity as individuals’ transition from adolescence into adulthood.
However, there is a need for a detailed understanding of physical activity behaviors in the
young adult population. It is possible that we can assess the transition between
adolescents and adulthood by studying college students, as they are a convenient
population to investigate. However, the literature on physical activity behaviors of
college students is limited and there is a need for more detailed assessments within this
population. Therefore, the focus of this literature review is to: 1) examine physical
activity patterns in college students, 2) exarrrine changes in physical activity patterns as
adolescents’ transition from high school to college, 3) examine the impact of enrollment
in activity related courses on physical activity levels, and 4) examine the changes in

physical activity participation that occur during college.

2.2 Physical Activity

Physical activity is a broad and complex concept that can be defined as any bodily
movement produced by skeletal muscles which result in a substantial increase total daily
energy expenditure '6. Under the broad concept of physical activity, it is important to
understand the differences between leisure time physical activity, exercise, sport, and

occupational physical activity. Physical activity that increases total daily energy

expenditure during an individual’s discretionary time is considered leisure time physical
activity. When leisure time physical activity is performed repeatedly over an extended
period of time with the intent to improve fitness, physical performance, or health, it is
often called exercise. Sport is a form of physical activity that involves competition
whereas, occupational activity refers to the energy expenditure required to meet the
demands of a job. Leisure time physical activity, exercise, sport, and occupational
activity are components of physical activity which is the only discretionary component of
total daily energy expenditure.

The relationship between physical activity and health is widely documented.
Physical activity and exercise can prevent the occurrence of cardiovascular events,
stroke, hypertension, diabetes, some cancers, obesity, depression, and can even delay
mortality. As a result of the many benefits of increased physical activity, an important
mission of the American College of Sports Medicine is to promote physical activity to the
general public through the generation and promotion of public health recommendations
for physical activity 4. In 1996, a Surgeon General’s Report titled Physical Activity and
Health 70 provided what are currently well-established physical activity
recommendations. This reports suggests that: 1) “Significant health benefits can be
obtained by including a moderate amount of physical activity on most if not all days of
the week.” and 2) “Additional health benefits can be gained through greater amounts of

physical activity.”

10

2.3 Physical Activity Assessment
Introduction

The accurate measurement of physical activity is fundamental to studies assessing
the relationship between physical activity and health. Physical activity energy
expenditure can be measured using a variety of indirect methods including doubly labeled
water 62 and accelerometry 59. Although these measures are often viewed as effective
methods for measuring energy expenditure, they are not suitable for studies including
large sample sizes. As a result, the most feasible method of collecting physical activity

information in a large population is via self-reported questionnaires.

Physical Activity Questionnaires

A global assessment of physical activity is often the primary concern when
assessing physical activity levels via questionnaire. However, physical activity
questionnaires often vary widely in detail. For example, questionnaires may assess
physical activity by categorizing individuals into dichotomous (active or inactive) or
multiple categories (sedentary, low active, moderately active, or highly active). Most
often, researchers use national recommendations for physical activity to create these
categories. Physical activity can also be assessed as a continuous variable such as energy
expenditure, which is often reported as the total amount of calories expended per day or
week during exercise 64. Self-report questionnaires require an individual to recall
physical activity levels over a specific time period ranging anywhere from the past day,
week, month, or throughout a lifetime. Because the reliability of self-reporting

information generally decreases over time, it is best to keep the reporting interval short 64.

11

Physical activity questionnaires focus on the type, intensity, frequency, and
duration of physical activity performed and typically focus on activities that are most
frequently performed 64. Intensity of exercise is most often reported in terms of
metabolic equivalents (METS). One MET is the resting metabolic rate while sitting or
lying quietly 2. Physical activity intensity is often expressed in multiples of METS and is
the intensity of an activity relative to one MET. For example, an activity with an intensity
of 3 METS is considered 3 times that of rest. The Compendium of Physical Activities 2
was developed to provide specific MET values for a variety of physical activities and is
often used to determine the intensity of physical activity in survey research. Frequency is
most often assessed by asking the number of times per week or times per month an
individual participates in a given activity. This is a reasonable approach in making an
overall assessment of physical activity 64. A second important aspect of frequency is
whether the activity is performed during a single session or split into several smaller parts
throughout the day. Thus, it may be important for surveys to distinguish between the two
by documenting the total number of sessions per week rather than the number of days per
week and individual participates in physical activity (’4. The duration of a single bout of
physical activity is often expressed in terms of minutes per session or minutes per day.
When duration is combined with the fi'equency of physical activity, total minutes of
physical activity can be determined. When the total minutes of physical activity is
combined with intensity of activity, energy expenditure can be calculated. Energy
expenditure can be reported as the total amount of calories expended per week during
exercise (kcal/week), the amount of calories expended per week during exercise relative

to body weight (kcal/kg/week), or total MET minutes per week. Assessing physical

12

activity in terms of energy expenditure provides the most detailed method of assessing

physical activity via questionnaires 64.

Reliability and Validity of Physical Activity Questionnaires

A major concern in physical activity survey research is the ability of a physical
activity questionnaire to accurately assess an individual’s physical activity behaviors. As
a result, previous research has focused on evaluating both the reliability and validity of
physical activity questionnaires. The reliability of a questionnaire reflects its ability to
yield the same result on multiple occasions. The validity of a physical activity
questionnaire reflects the degree in which the survey measures the physical activity
parameter it proposes. This section was limited to the review studies assessing the
reliability and validity of commonly used physical activity questionnaires. A total of 11
studies assessing reliability and 6 studies assessing validity were reviewed.

Reliability: The reliability of physical activity questionnaires is most often
assessed using test-retest intraclass correlation coefficients which measure the degree to
which two surveys administered on the same individual agree with each other. The
reliability of two versions of the International Physical Activity Questionnaire (IPAQ)
was assessed in adults between the ages of 18 and 65 across multiple countries. The short
version of the IPAQ was designed to assess time spent walking, in vigorous activity,
moderate activity, and sedentary activity over the previous seven days. The long version
of the IPAQ was developed to collect more detailed information on the type of physical
activity performed. Exercise intensity was assessed using MET values based on the

Compendium of Physical Activities and total physical activity was reported in MET-

l3

minutes per week. Data from 1880 adults across 12 countries were used to assess the
test-retest reliability in both the long and Short forms of the IPAQ questionnaire. The
one-week, test-retest reliability data were comparable between both forms of the IPAQ
questionnaires showed a Pearson correlation coefficient (r) of nearly .80 in adults who
completed the questionnaire '7 indicating good reliability.

Lamb and Brodie 38 assessed the test-retest reliability of Liverpool Leisure-time
Physical Activity Questionnaire (LLTPAQ) in 62 male and female British university staff
and students. Total leisure time physical activity was assessed in terms of caloric
expenditure by asking subjects to report activities performed over the past 14 days, the
duration of each activity, and if sweating occurred during the activity. The test-retest
reliability of the survey was assessed by inviting the subjects to return for a repeat survey
4 weeks later. The four-week test-retest reliability of total leisure time physical activity
was r = .86 implying that the LLTPAQ is an excellent tool for recording physical activity.

The Minnesota Leisure Time Questionnaire is another survey instrument also
used to assess leisure time physical activity in adults. From a list of 63 different physical
activities, subjects were asked to report activities performed over the prior 12 months.
Energy expenditure was then calculated by asking subjects to report the number of
months they participated in each activity, the average number of sessions per month, and
the average time per session that was spent participating in that activity. The
approximate calories per day of leisure time activity were then calculated. The test-retest
reliability of this questionnaire was assessed in 147 adults over a five week period.
Within this study population, the test-retest correlation coefficient for total leisure time

physical activity was .88 indicating excellent reliability 22.

l4

The Physical Activity Index from the College Alumnus Questionnaire was
designed to estimate caloric expenditure over a one-week period while climbing stairs,
walking, participating in sports, and participating in recreational activities. In order to
assess the test-retest reliability of the questionnaire, the survey was repeated one month,
eight months, and nine months following its administration at baseline. The test-retest
reliability was high when measured at 1 month (r = .72), however, when assessed at 8 to
9 months, the test-retest reliability dropped substantially to r=.34 and r=.43, respectively.
These results illustrate that the reliability of this instrument diminished as the recall
period lengthened 3.

In order to offer a more global perspective of reliability, a simultaneous
evaluation of the reliability of multiple physical activity questionnaires was assessed in
78 subjects varying in age from 20-59 33. The test-retest correlation coefficients over a
one-month period exceeded .92 for total physical activity in the Minnesota Leisure Time
questionnaire, the Lipid Research Clinic questionnaire, and the Baecke Physical activity
questionnaire. The one-month test-retest reliability for the Godin Leisure time physical
activity questionnaire (.62), the health insurance plan of New York (.86), The CARDIA
physical activity history questionnaire (.88), the College Alumnus questionnaire (.72), the
Minnesota heart health program (.86), and the Stanford usual activity questionnaire (.77)
showed moderate to high reliability. These results illustrate that most physical activity
questionnaires are reliable over a short time period.

Booth and colleagues 7 assessed the reliability of the World Health Organization
health behavior in schoolchildren survey. The questionnaire was administered on two

separate occasions, two weeks apart and the test-retest reliability was assessed by

15

measuring percentage of agreement rather than using the preferred method of correlation
coefficients. Agreement was calculated as the number of students classified as active at
both occasions plus the number of students classified as inactive at both occasions,
divided by the total population. When frequency and duration of physical activity were
assessed together, there was 67% agreement among 14 year old boys, 70% agreement
among 14 year old girls, 85% agreement among 16 year old boys, and 70% agreement
among 16 year old girls. The agreement shown between the two-week time period in this
study is considered substantial 39.

Validity: The validity of a questionnaire reflects the degree to which physical
activity questionnaires measure what it is propose to measure. Compared to reliability,
validity is assessed less often due to the absence of a widely accepted criterion measure
of physical activity 64. Despite this, validity remains a critically important issue in the
assessment of physical activity questionnaires. Most often, the validity of physical
activity questionnaires is determined by comparing the questionnaire to another measure
of physical activity such as doubly labeled water or accelerometry.

The doubly-labeled water technique is particularly useful for measuring energy
expenditure over relatively long periods of time (a few days or weeks). However, the
validation of a physical activity questionnaire against double labeled water is costly and
as a result most validation studies include only small sample sizes. The validity of three
physical activity questionnaires was compared to physical activity levels measured by
doubly labeled water in 19, 40-year-old males 55. The Baecke physical activity
questionnaire assessed four indices of physical activity: work index, sport index, leisure

time index, and the total activity index. The Five City Project questionnaire assessed

l6

physical activity levels by asking the frequency a sweating during vigorous activity in
combination with the number of hours of vigorous activity per week. The Tecumseh
Community Health Study questionnaire assessed daily energy expenditure by assessing
the number of hours spent in various physical activities per week. The Baecke Activity
Questionnaire showed the highest correlation with doubly labeled water (r=.69). Total
daily energy expenditure measured by the Tecumseh Community Health Study
questionnaire also showed a high correlation (r=.64) with double labeled water. There
was also a good correlation (r=.5 7) between the sweat index measured by the Five City
Project questionnaire and double labeled water 55. Furthermore, when compared to
doubly labeled water, the validation of the Physical Activity Scale for the Elderly showed
similar correlation coefficients for both males (.79) and females (.68) 63. The results of
these previous studies suggest that physical activity questionnaires show good validity
when compared to doubly labeled water.

Validation studies performed in larger samples often use an accelerometer
because it is a less expensive method to assess physical activity. In fact, the International
Physical Activity Questionnaire was assessed against accelerometry in 744 adults across
12 countries. Correlation coefficients were calculated to assess the validity and the results
illustrated only a fair to moderate agreement between the two measures (r = .33) '7.
Philippaerts and colleagues 54 also assessed the validity of two physical activity
questionnaires, the Tecumseh Community Health Study Questionnaire and the Beacke
Activity Questionnaire, using accelerometry as the criterion variable. Accelerometer data

were collected over four consecutive days in 166, 40 year old men. Both the Tecumseh

l7

Community Health Study Questionnaire and the Beacke Activity Questionnaire showed a
moderate validity (r = .47 and r = .32, respectively) compared to accelerometer data.
Physical activity diaries are also used as an additional method of validating
physical activity questionnaires. The College Alumnus questionnaire was validated
against a 48 hour physical activity diary. Seventy eight adults between the ages of 21 and
59 recorded all physical activity at least every four hours in a diary designed for the
study. Following the 48 hour period, the subjects completed the College Alumnus
Questionnaire. Correlation coefficients between the two measures were calculated and
results showed that the physical activity index of the College Alumnus Questionnaire
moderately agreed with the physical activity diary (r = .42) 3. Aadahl and Jorgensen '
assessed the validation of a physical activity questionnaire for measuring sport, work, and
leisure time activities against a 4-day physical activity diary. Using the Compendium of
physical activities, total MET time was calculated by assessing the number of hours and
minutes spent on each MET activity on an average weekday. The validity of the physical
activity questionnaire was assessed in 20 men and 19 women. Within the diary,
participants were requested to report the type, duration, and intensity of all activities, 24
hours a day, for a four day period. The agreement between the activity questionnaire and
the diary was then assessed and the results illustrated the correlation between physical
activity questionnaire and the diary was high (r = .74) indicating that this self-report

physical activity questionnaire was valid.

18

Conclusion

The ability of a questionnaire to: 1) yield the same result on multiple occasions
and 2) to measure what it is proposes is fundamental to the accurate assessment of
physical activity. Thus, previous research has examined both the reliability and validity
of a variety of physical activity survey instruments. Previous studies illustrate the test-
retest correlation coefficients of well-known physical activity surveys range from .72-.92
over a one month period. These results suggest that questionnaires assessing physical
activity are reliable over a short time period. The validity of physical activity
questionnaires has been assessed against a variety of physical activity measures. As a
result, correlation coefficients have ranged from .32 to .79 in studies using doubly labeled
water, accelerometry, or physical activity diary as the comparison measure. Despite the
wide range of correlation coefficients, there appears to be a moderate to high relationship
when comparing questionnaires to other forms of physical activity measures. These
results illustrate that questionnaires are a sufficiently accurate method for assessing

physical activity in population based studies.

2.4 Physical Activity in College Students
Introduction

Recently, the American College Health established national college health
objectives in a document titled “Healthy Campus 2010: Making It Happen” 5. Healthy
Campus 2010 is a companion document to the healthy People 2010 and serves as a basis
for developing plans to improve student health. Healthy Campus 2010 has developed

over 200 health objectives with baselines and targets for the nation's colleges and

19

universities. Physical activity is one of the top ten leading health indicators within this
document. The two specific physical activity objectives of Healthy Campus 2010
include: 1) increasing the proportion of college students who have received information
on physical activity and fitness and 2) increasing the proportion of college students who
engage in physical activity at least 3 days/wk at moderate intensity for at least 30
minutes, or vigorous physical activity for 20 minutes or more minutes. In order to
determine whether Healthy Campus 2010 goals are met, detailed assessments of college

students’ physical activity levels are needed.

Review Studies of College Students ’ Physical Activity Levels

Despite the Healthy Campus 2010 goal to improve physical activity levels among
college students, the current literature to assess physical activity behaviors in college
students is inadequate. This point has been illustrated in two systematic reviews
conducted specifically to interpret the previous literature assessing physical activity in
college students. To the best of our knowledge, these are the most thorough studies that
have been performed to assess physical activity participation in college students, and the
purpose of this section is to describe their results.

Recently, a review article on physical activity behaviors in college students
addressed three key issues: 1) previous research on students’ physical activity levels, 2)
problems surrounding current research on this topic, and 3) suggestions for further
research 36. This review illustrated that a glaring limitation among studies assessing
college students’ physical activity levels is inconsistent measures used to assess physical

activity. Currently, there is no consensus regarding the use of standardized questions to

20

assess physical activity in a college population. Although energy expenditure is
calculated using intensity, total time, and frequency of activity, the studies assessed in the
review used various combination of the three including: 1) frequency and intensity of
specific activities, 2) total weekly minutes of physical activity regardless of intensity, and
3) only frequency of physical activity. The authors concluded that these inconsistencies
in t physical activity measurement makes it nearly impossible to summarize current
physical activity patterns in this population 36.

In order to assess the prevalence of university students meeting the American
College of Sports Medicine/Centers for Disease Control and Prevention (ACSM/CDC)
guidelines for moderate activity (students who accumulated 30 minutes or more of
moderate physical activity on most, preferably all, days of the week), a systematic review
including 19 studies representing nearly 36,000 students from a total of 27 countries was
performed 32. Results Showed that approximately 50% of university students within the
United States, Canada, and China did not meet the recommendations for moderate
physical activity. Within Europe and Australia, the prevalence for not meeting physical
activity recommendations was 67% and 40%, respectively. However, the authors
cautioned that lack of a standard method of physical activity assessment is a major
limitation when assessing those data. Two general conclusions were drawn fiom the
results of this study. First, insufficient physical activity is a serious health problem
among university students. Second, there is a need for a standardized tool to assess
physical activity within college students to better track their physical activity

participation 32.

21

In conclusion, the two systematic reviews assessing physical activity participation
in college students illustrate similar methodological limitations within the previous
research. These limitations include the inconsistent measurement of frequency, intensity,
and duration of physical activity participation. Although these studies conclude that
physical inactivity is a problem within the college population, researchers should strive to
assess physical activity participation in a more consistent manner. This will offer a more

detailed and consistent perspective of physical activity behaviors within this population.

Physical Activity Participation in College Students

Healthy Campus 2010 included physical activity as a priority health risk behavior
based on its relevance as a broad public health issue 5 . As a result, it is necessary to
determine the proportion of the general college student population who participate in
adequate amounts of physical activity and to determine what characterizes physical
activity participation in this population. The search terms “physical activity”,
“exercise”, “college”, “university”, and “student” were used in various combinations to
conduct the literature search via the National Library of Medicine to identify all relevant
articles. This was followed by a careful examination of literature cited pages to identify
additional relevant studies that were overlooked. To our best knowledge, twelve studies
have assessed physical activity participation in a general population of college students,
all of which will be reviewed in this section. In addition, this section will be organized
by the method of physical activity measurement. First, the six studies reviewed in this

section assessed physical activity participation as meeting the ACSM/CDC national

recommendations for physical activity. It should be noted that these recommendations

22

for moderate physical activity (30 minutes per day, five or more days per week) are
different that the Healthy Campus 2010 goals for moderate physical activity (30 minutes
per day, three or more days per week). To the best of our knowledge, no previous studies
the criteria of Healthy Campus 2010 to assess moderate physical activity. Second, the
Single study that estimated total caloric expenditure was reviewed separately. Finally, the
remaining five studies that used various combinations of frequency, intensity, and
duration of physical activity were grouped.

Meeting the national recommendations for physical activity: The American
College of Sports Medicine recommends that every American adult accumulate at least
30 minutes of moderate physical activity on most days of the week or 20 minutes of
vigorous physical activity on 3 days per week 4. When assessing recommendations,
moderate activity is most frequently defined as participating in an activity such as
walking, slow biking, or gardening that did not make you sweat or breathe hard.
Vigorous activity is most fiequently defined as participating in an activity such as
jogging, running, rowing or other activities that made you sweat and breath hard. These
recommendations are often used to assess college student physical activity participation.

In order to monitor health risk behaviors among high school youth in the United
States the Youth Risk Behavior Surveillance System (YRBSS) was developed. A
component of the YRBSS is the Youth Risk Behavior Survey (YRBS) which assesses a
variety of health related behaviors including physical activity participation in high school
students. An additional component of the YRBSS is the National College Health Risk
Behavior Survey (N CHRBS) which focused specifically on health risk behaviors within

the college population. The NCHRBS was administered in 1995 and used a nationally

23

representative sample of undergraduate college students fi'om 136 universities across the
United States. A random sample of students was drawn from a list of undergraduate
students older than 18 years. To ensure adequate proportions of minority students,
differential sampling rates were used. A total of 8810 students were selected from the
136 colleges and universities and data fiom 4609 undergraduates were available for
analyses (60% response rate). Of the 4609 college students, 64% were between the ages
of 18 and 24. The results showed that 42% of students between the ages of 18-24
participate in adequate levels of vigorous activity while 21% participate in adequate
amounts of moderate physical activity 48. In addition, NCHRBS results indicated that
male students (49%) were more likely to report vigorous exercise than female students
(35%) and that black students (28%) were significantly more likely to report moderate
activity levels than white students (18%).

Physical activity participation during college was assessed as part of a
cardiovascular needs assessment in 607 first year students enrolled in a Nova Scotia
University 44. Results showed that 51% of the students did not participate in adequate
amounts of vigorous physical activity. Of students who reported any physical activity,
the most frequently reported mode of activity was walking (64%). In addition, there was
a positive relationship between perceived knowledge of cardiovascular benefits of
physical activity and frequency of physical activity participation, which illustrate the
importance of increasing physical activity knowledge in college students 44.

Gyurcsik and colleagues 26 assessed vigorous physical activity participation in
132 female freshmen enrolled in their first-year of college immediately following high

school graduation. Participants were asked to indicate the number of times they engaged

24

in at least 20 minutes of vigorous physical activity during a typical week over the
previous month. The average frequency of vigorous physical activity was only 2.8
weekly sessions and 47% failed to meet the national recommended level of vigorous
physical activity 26.

Vigorous physical activity participation during the first two months of college
was assessed in 145 first year college students attending a liberal arts university in
Canada ”. The students who reported participating in an average of three or more bouts
of vigorous activity for 20 minutes or longer per week during the first two months of
college were classified as active. Forty four percent of the college population was active
during the first two months of college ”.

It is possible that the setting in which one exercises has an effect on the meeting
the recommendations for moderate intensity physical activity. In order to assess the
impact of exercise setting, self-reported physical activity was assessed in 594 first and
second year university students '3 . Students were asked to identify the frequency,
duration, and intensity of physical activity when participating in four different exercise
settings: 1) aerobics classes, 2) with others outside of a structured aerobic class, 3) alone
in an exercise setting, and 4) completely alone. Six percent of the population reported no
involvement in physical activity within the four listed settings. Of those students who
reported involvement in a physical activity setting, the average frequency and duration of
exercise was 4.5 days per week and 52 minutes, respectively. Within the total sample,
43% met the national recommendations for vigorous activity. Only a small percentage of
students (10%) who met the recommendations were physically active in only one

exercise setting, whereas 28% obtain their activity in at least two contexts and 62%

25

obtain their activity in three or more contexts. When compared to the perecentage (10%)
of students who met the physical activity recommendation, a larger percentage (30%) of
active students who did not meet these recommendations obtained their activity in only
one settings (30%). In addition, only 33% of students who did not meet the
recommendations exercised in three or more settings compared to 62% in those who did
meet the recommendations. These results suggest that exercising in multiple settings are
associated with improved adherence to the ACSM/CDC national physical activity
recommendations 13.

Pinto and colleagues 57 examined self reported physical activity data in 332 first-
year college students enrolled in a private university. Students were divided into two
groups. Students who reported either: 1) 20 minutes of vigorous physical activity
(exercise that made you sweat and breath hard) three or more days per week, or 2) 30
minutes of moderate physical activity (exercise that did not make you sweat and breathe
hard) five or more days per week were met the recommendations and were classified as
“active”. Results indicated that 58% of the sample exercised at or above the
recommended level and 42% did not 57.

In summary, these study results suggest that a substantial amount of college
students do not meet the current physical activity recommendations. Inadequate levels of
physical activity ranged from 42% to 56% in these studies. In order to track
improvements, researchers should continue to assess college students’ physical activity in
terms of meeting the national recommendations for both vigorous and moderate activity.

Physical activity caloric expenditure in college students: To provide a more

detailed assessment of physical activity via questionnaire, caloric expenditure can be

26

calculated using data gathered from questions assessing the frequency, intensity, and
duration of physical activity. To our knowledge, only one study reported physical
activity levels in terms of caloric expenditure in a general population of college students.

Leslie and associates 43 identified insufficient physical activity among 2,729
Australian students enrolled in two metropolitan and two rural universities. A self
reported 2-week physical activity recall was used to assess the fi'equency and duration of
walking for recreation, moderate physical activity, and vigorous physical activity.
Metabolic equivalents were used to estimate energy expenditure over a 2-week period.
Within the male student population, 8% were classified as sedentary (< 100 kcal/week),
24% were low-active (100-799 kcal/week), 42% were moderately active (>800
kcal/week, but not in vigorous category), and 26% were highly active (>1600 kcal/week
including three sessions of at least 20 minutes of vigorous activity). Eleven percent of
females were sedentary, 36% were low-active, 36% were moderately active, and 16%
were highly active. This analysis was extended to examine associations of personal,
social, and perceived environmental variables in relation to physical activity status. In
comparison to less active students, sufficiently active college students were more aware
of campus recreational facilities, had a high enjoyment of physical activity, high social
support fi'om family and friends, high self-efficacy, were employed, and were members
of a gym.

Physical activity in college students using various methodologies: As referenced
by two previous review studies 32’ 36, the biggest limitation in evaluating physical activity
in college students is the inconsistent methodology used to assess physical activity. The

purpose of this section is to review studies that used various combinations of fi'equency,

27

intensity, and duration of physical activity to assess physical activity participation in
college students.

A large cross-sectional survey assessing leisure time physical activity in over
19,000 college students across 23 countries was performed in 2004 to analyze differences
in physical activity prevalence. Leisure time physical activity was assessed by asking the
type and frequency of physical activity performed over the past two weeks. Subjects
were divided into three groups: inactive (0 days of exercise per week), low-frequency
activity (1-2 days of exercise per week), and recommended frequency activity (exercising
three or more times per week). Overall, more women (3 8%) then men (27%) reported no
leisure time physical activity. When comparing countries, inactivity levels were lowest
in Western Europe and the United States for both men (24%) and women (21%). The
highest rates of physical inactivity were seen in developing countries which included
Colombia, South Afiica, and Venezuela (35% in men, 53% in women) 27. A major
limitation to this study was that intensity and duration of physical activity was not used to
describe physical activity levels.

Sparling and Snow 65 assessed physical activity levels in recent college alumni. A
survey was administered to 376 individuals who recently graduated from college.
Physical activity during college was assessed retrospectively by asking the alumni to
report their physical activity patterns during their final year of college. Nearly 43% of
the sample considered themselves to be regular exercisers (3 or more days of physical
activity per week) during college, 40% as irregular exercisers (1-2 days of physical
activity per week), and the remaining 17% as non-exercisers (less than one day of

exercise per week). The proportion of men who categorized themselves as regular

28

exercisers was 45% compared to 39% of women. Interestingly, 85% of study
participants who were regular exercisers during college were either as active or more
active following graduation and 81% of non-exercisers reported being the same or even
less active following graduation. These results suggest few individuals changed there
physical activity participation soon after graduation.

Self-reported participation in physical activity was analyzed in 217 college
students attending a private, liberal arts university in Rhode Island. Physical activity data
were obtained fi'om undergraduate students completing a survey assessing satisfaction of
university health care service. Both type and frequency of physical activity were
assessed and individuals who exercised three or more times per week were considered
active. The results illustrated that 54% of the population was currently active, whereas
28% exercised up to two times per week. The remaining 18% were considered inactive.
There were no significant differences between gender and class across the three groups.
Furthermore, students reported participating in running, weight lifting, and cycling most
often 58.

Dinger and Waigandt '9 collected physical activity data on over 2600 college
students enrolled in a large Midwestern university. F orty-five percent of the sample
reported participating in at least 3 days of vigorous activity during the previous week
while 22% did not report participating in vigorous activity. F orty-six percent of the
population participated in moderate physical activity three or more days per week.
However 30% of the sample did not report participating in moderate physical activity the
previous week. On average, males participated in an average of 2.8 days of vigorous

physical activity per week, 2.4 days of moderate activity per week, 1.9 days of flexibility

29

exercises per week, and 1.8 days of muscular strengthening exercise per week. In
comparison, females participated in an average of 2.2 days of vigorous physical activity
per week, 2.9 days of moderate activity per week, 2.2 days of flexibility exercises per
week, and 1.5 days of muscular strengthening exercise per week.

Inappropriately high levels of physical activity in college students has also been
examined. A study by Garman and colleagues 25 examined the occurrence of excessive
exercising in a college population. A total of 268 students completed a self-report
questionnaire designed to assess frequency, duration, and intensity of physical activity.
Participants were considered to be “exercise dependant” if they: 1) exceeded 360 minutes
of exercise per week and 2) pursued physical activity when it could compromise health or
supersede social and/or academic responsibilities. Of the 22% of the sample who were
considered exercise dependent, 53% were male. The average time spent exercising
within the exercise dependent group exceeded 730 minutes per week. Within the
population of college students who were considered general exercisers (students
participating in less than 360 minutes of exercise), an average of 103 minutes of physical
activity was performed per week. The average frequency of physical activity
participation was just under three times per week in the general student population.

In summary, a variety of different methods have been used to assess physical
activity participation in college students. Although varying methodologies make direct
comparison of results between previous studies difficult, the majority of findings indicate
a large proportion of these students are not physically active. Therefore, future research
should include detailed assessments of physical activity to better describe the current

physical activity insufficiencies within the college population.

30

Physical Activity Levels among College Students Enrolled in a Health Related Class

Much of the physical activity research in college students is collected when they
are enrolled in health related courses. Thus, a potential limitation is that these data may
not be representative of the general university population. It is possible that college
students who are more physically active and/or health conscious will enroll in health
related courses due to their interest in the topic. As a result, the purpose of this section is
to review the physical activity literature in college students enrolled in a health related
course separately from the studies conducted in the general student population. To our
knowledge, a total of seven studies assessing physical activity levels in college students
enrolled in an activity based class, all of which were included in this section.

Project GRAD (Graduate Ready for Activity Daily) '4’ 6’ was developed to
promote the adoption and maintenance of physical activity among young adults
transitioning from the university into adult roles. Three hundred and thirty eight college
seniors from a large urban university were randomized into a course designed to promote
physical activity or a control course which covered general health topics. Each subject
completed a 7-day physical activity recall which assessed total energy expenditure,
minutes per week of moderate exercise, and minutes per week of vigorous exercise at the
beginning and end of the 15-week classroom based intervention. The classroom based
intervention had no significant effects on total leisure time physical activity participation
in men. However, women showed improvement in total leisure time physical activity,
flexibility exercises, and strengthening exercises 6'. In addition, there were no significant
intervention effects on physical activity in either men or women two years following the

intervention '4.

31

Anding and colleagues 6 assessed physical activity levels in a convenience sample
of 60 college women. Study participants were enrolled in one of three aerobics courses at
a university which requires students to complete one semester of an activity related
course. Students completed the Self Report Physical Activity Scale (SRPA) to indicate
their physical activity level during the past month on a scale from 0 (avoiding physical
activity entirely) to 7 (exercising more than 10 hours per week). Based on the results of
the SRPA, only 33% of the sample was considered physically active. However, because
these study participants were enrolled in an aerobics course, it is possible they wanted to
increase their physical activity levels.

Physical activity, fitness, body composition, muscular strength, and flexibility
were assessed in 115 male and 143 female college students enrolled in a general wellness
course emphasizing the importance of physical activity and fitness 56. Physical activity
was evaluated via questionnaire by classifying activities into one of three categories: 1)
aerobic, 2) recreational, and 3) weight training. Students who participated in at least two
sessions of a category per week were considered a participant within that category. The
results for physical activity patterns indicated the proportion of females likely to
participate in aerobic training (78%) was significantly higher than males (34%).
However, a significantly greater proportion of males (34%) were more likely to
participate in weight training than females (18%). In addition, males (28%) were also
more likely to participate in recreational activities than females (12%).

A total of 274 male and female college students enrolled in a general health
course were asked to complete a survey regarding their personal health choices. Physical

activity participation was included in the assessment 5 1. Results of the study illustrated

32

that only 40% of males and 32% of females reported maintaining a regular physical
activity schedule. Study participants were also asked to report the number of days in
which physical activity participation was sufficiently intense to significantly increase
heart rate for 20 minutes or more. Males reported significantly more days (3.9) of
exercise that increased heart rate compared to females (3 .2). The authors concluded that
a large proportion of college students, especially females, are not physically active.

Brevard and associates '2 examined the impact of residence on physical activity in
104 college students enrolled in an introductory nutrition class. A self-report
questionnaire was used to assess both residence and physical activity within this sample.
Residence was assessed by categorizing subjects as either living on-campus or living off-
campus. Physical activity levels (sedentary, lightly active, active, very active), mode
(aerobic, anaerobic, combination, no exercise), and weekly caloric expenditure (0, 1-399,
400-899, 900-1499, 100-2499, >2500 kcal/week) were determined for each subject.
Results of the study showed that activity level and mode of exercise did not differ
between students living on-campus and off-campus. Average weekly caloric expenditure
was lower in males living on-campus (4881 kcal/week) when compared to males living
off-campus (6281 kcal/week). In contrast, average weekly caloric expenditure was
higher in women living on-campus (2173 kcal/week) compared to those living off-
campus (1420 kcal/week). However, these differences were not significant.

The majority of physical activity data in college students has been collected in
predominantly Caucasian samples and is therefore may not be generalizable to the
African American college student population. Ford and Goode 23 assessed daily physical

activity levels in African American college students. Study Participant included 224

33

undergraduate students enrolled in a health education class at a predominantly African
American institution. Nearly 55% of the sample did not participate in daily physical
activity. Of the 45% of the sample that participated in daily physical activity, 60% were
males.

Physical activity levels were examined in 254 African American college students
enrolled in a personal health class 37. Physical activity was assessed via the Lipid
Research Clinics Physical Activity Questionnaire (LRCPAQ) which assessed physical
activity during work, physical activity outside of work, and the frequency of participation
in strenuous activity. Based on a combination of responses, subjects were classified as
very low active, low active, moderately active, or high active. Results showed that
physical inactivity was significantly different between males and females. In fact,
according to the LRCPAQ classifications, 65% of females were classified as either low
active or vary low active compared to 42% in males. The results suggest a large
proportion of African American college students, especially females, are not physically
active 37. In addition, study participants were asked to complete the Physical Activity
History Questionnaire (PAHQ) which assesses participation in 13 specific vigorous and
moderate intensity activities. Results fi'om the PAHQ suggested males participate most
frequently in: 1) high intensity home and leisure activities such as weight lifting and 2)
strenuous sports such as basketball, football, and skiing. For moderate intensity
activities, males were most likely to report participation in walking and non-strenuous
sports such as softball, volleyball, and swimming. Within the female population, rtmning
and exercise dancing were reported most frequently for high intensity activities while

walking and home exercises were reported most frequently for moderate activities.

34

These results offer a more detailed perspective on the type of physical activity among
African American college students 37.

To our knowledge, these are the only two studies which have focused only on
Afi'ican American students. These results illustrated that 42-65% of all African
American college students were physically inactive, which is similar to the 42% to 56%
26’ 32‘ 44’ 48’ 57 that found in the general college student population. To firrther assess ethic
disparities in physical activity levels in college students, Suminski and colleagues 67
reported that physical activity participation (global self-rating of physical activity
participation during the previous month) did not differ between Afiican Americans and
Caucasians. Therefore, the current literature suggests that that physical activity
participation in African American and Caucasian college students does not differ.

As stated earlier, it is possible that a more health-conscious, physically active
college student may be more likely to enroll in activity related course. As a result, their
physical activity levels may be different from the general population of college students.
However, no previous study has directly addressed whether this selection bias exists.
Results from previous studies illustrate that the proportion of inactive college students
enrolled in a health related course ranged from 42% to 77% which is similar to the 42%
to 57% that found in the general college student population. Although these results
suggest that physical activity may not differ between students who choose to enroll in an
activity related course and those who do not, there is a need to assess this potential

selection bias in more detail.

35

2.5 Decline in Physical Activity from High School into College
Introduction

The decline in physical activity with age is one of the most consistent findings in
physical activity epidemiology 60. Childhood is the most physically active time during an
individual’s life 63* 73. However, it appears the decline in physical activity begins during
adolescence and continues to decline throughout a lifespan. Although it is accepted that
physical activity declines with age, the reasons for this trend are not well understood 60.
As a result, there is a need for a more detailed understanding of physical activity
behaviors across a variety of age groups including young adults as they transition from
high school to the university setting.

The transition from high school to university is a complex phenomenon that has
been considered more than a single event 24. This process appears to be a unique
experience to each individual and is of substantial interest to educational professionals.
The changes that occur during the transition fi'om high school to university can be
academic, social, cultural, emotional, and physical 24. The purpose of this section of the
literature review is to focus on this transition and assess changes in physical activity
levels between high school and college. Currently, there is a lack of research on the
change in physical activity from high school to college. The terms high school, college,
university, physical activity, sport, exercise, adolescent, young adult, and transition were
used in various combinations when searching the National Library of Medicine database.
Following identification of all relevant articles, a careful examination of each literature
review was conducted to identify any additional relevant studies that may have been

overlooked. Our search yielded a total of twelve relevant studies which are included in

36

this section of the literature review. Two large population based studies were used to
compare differences in the prevalence of physical activity in high school students and
college students. To the best of our knowledge, only four studies focused Specifically on
the Specific transition between high school and college while eight studies focused on

longitudinal changes in physical activity from adolescence into young adulthood.

Physical Activity during the Transition from High School to College

The concern that physical activity declines fi'om high school to university is
illustrated by comparing prevalence data between the National College Heath Risk
Behavior Survey (N CHRBS) and the Youth Risk Behavior Survey (YRBS). In 1995,
surveillance data on health risk behaviors of college students were collected as part of the
NCHRBS. This survey showed that only 38% of university students participated in
vigorous physical activity on three or more of the past seven days 20. Compared to the
NCHRBS, the prevalence of vigorous physical activity was much higher (55%) in high
school students who were part of the 1995 YRBS 35. These findings provide evidence of
a decrease in physical activity as a student transitions from high school to the university.
However, these results are limited in that data are derived fiom two separate cross-
sectional populations using different survey techniques and methods. AS a result, there is
a need for longitudinal data on physical activity levels during the transition from high
school to college.

In order to assess this transition period, Bray and Born ” compared vigorous
physical activity levels between high school the first year of college in 145, 18 and 19

year old students enrolled in a university. High school physical activity levels were

37

assessed retrospectively. Participants who reported an average of three or more bouts of
vigorous physical activity (any physical activity strenuous enough to cause one to break a
sweat or breathe heavy) lasting 20 minutes or longer per week during the last 8 weeks of
high school were classified as active. Similarly, those who participated in three or more
bouts of vigorous physical activity lasting 20 minutes or longer per week during the first
8 weeks of college were classified as active. Individuals who reported less than three
sessions of vigorous exercise lasting 20 minutes or longer were classified as insufficiently
active. Results of the study showed that fi'equency of physical activity declined from 3.3
sessions per week during high school to 2.7 sessions per week during college. Sixty six
percent of the population was classified as active during the last two months of high
school. The proportion of active students significantly dropped to 44% during the first 2
months of college. Furthermore, 33% of respondents were active continuously (active
during high school and college), 23% were insufficiently active continuously
(insufficiently activity during high school and college), 11% were insufficiently active
during high school and became active during college, while 33% were active during high
school and became inactive during college ”. These results illustrate a decline in
physical activity as student’s transition from high school to the university.

Bray '0 also examined barriers to physical activity as a predictor of physical
activity levels during the transition form high school to college. A total of 127 first year
students at a liberal arts university in Canada provided self reports of their physical
activity for the eight months prior to enrolment in the university and the first year at the
university. Results showed that physical activity tracked moderately (r=.58) between

high school and college. Furthermore, the ability of college students to cope with

38

potential barriers to physical activity (lack of motivation, cold weather, demands both
inside and outside of school, social barriers, etc.) predicted physical activity levels during
this transition '0.

To examine further the changes in physical activity participation during the
transition from high school to college, 149 female and 46 male first-year university
students enrolled in a health science course completed 7 day recall of vigorous physical
activity 40. Average weekly caloric expenditure for the entire p0pulation was 27.8
kcal/kg/week and there were no significant gender differences. When asked about
previous physical activity levels, 42% believed they were less active now compared to
their last year of high school, 25% thought they performed similar amounts of vigorous
physical activity, and 33% thought they engaged in more vigorous physical activity 40.

Everhart and colleagues 2' assessed the impact of sport participation and physical
education enrolhnent during high school on physical activity levels during college.
Current physical activity levels were assessed via questionnaire in 201 college students
enrolled in two southwestern universities. Number of days per week students
participated in at least 30 minutes of moderate-to-vigorous physical activity, strength
training, cardiovascular activities, team sports, aquatic participation, and abdominal
exercise were assessed separately. Results illustrated that students enrolled in high
school physical education did not report higher physical activity levels during college,
but did report increased strength training participation. Individuals participating in high
school athletics, with or without high school physical education, reported greater team
sports participation and strength training during college than non-athletes enrolled in

physical education 2'.

39

In conclusion, there are minimal data comparing physical activity behaviors
between high school and college students. Physical activity appears to decline, but there
is a need for a more detailed perspective on the changes in physical activity during this
time period. Researchers should focus on changes in sport participation, leisure time

participation, and caloric expenditure as students move from high school to college.

Changes in Physical Activity between Adolescence and Young Adulthood

As stated previously, there is minimal research that focuses specifically on the
changes in physical activity during high school and college. The majority of studies
focus on the overall decline in physical activity between adolescence and young
adulthood. Caspersen et al. 15 examined physical activity patterns across adolescents,
young adults, middle aged adults, and older adults. Data fi'om over 10,000 males and
females between the ages of 12 and 21 participating in the 1992 Youth Risk Behavior
Survey and over 43,000 males and females greater than 18 years participating in the 1991
National Health Interview Survey were used to assess physical inactivity, light to
moderate activity, and vigorous activity. The prevalence of adolescent physical inactivity
(no participation in vigorous or moderate physical activity) increased fi'om 6% at age 14
to 24% at age 20 for male and female subjects. The prevalence of reporting regular,
sustained physical activity dropped from 40% to 24% in boys and fi'om 30% to 20% in
girls between the ages of 12 and 17. Between the ages of 12 and 21, vigorous physical
activity declined from 76% to 42% in males and from 66% to 28% in females and

continued to decline during young adulthood. This was followed by a more stable pattern

40

of physical activity prevalence between the ages of 30 and 64. In conclusion, physical
activity behaviors decrease throughout adolescence and into young adulthood.

The Cardiovascular Risk in Young Finns Study is a longitudinal project
developed to assess physical activity patterns over a 21 year period from childhood to
adulthood in Finland. Baseline measures were gathered in 1980 on 2309 children and

8 69. A physical activity index was calculated by

adolescents between the ages of 9 and 1
assessing, the fi'equency and intensity of leisure time physical activity, participation in
sport club training, participation in competitive sports events, and participation in
physical education class. Continuously active subjects were defined as those who were in
the highest tertile of the physical activity index score over a three year period (1980 to
1983) and a six-year period (1980 to 1986). Continuously inactive subjects were defined
as those who were in the lowest tertile of the physical activity index over the same time
period. The outcome variable was membership in the highest tertile 21 years following
baseline measures. Logistic regression analysis indicated that 15 and 18 year old males
who were active at baseline were more likely to be physically active 21 years later (OR:
5.7). The odds ratios for male subjects who were continuously activity over a three year
period compared to those who were continuously inactive was 11.8 while the odds ratios
for subjects who were continuously active over a 6 year period was 19.2. For females
who were ages 15 and 18 years at baseline, the odds ratio of being physically active 21
years later was 2.8 if active at baseline, 4.4 if continuously active over a three year period
and 6.1 if continuously active over a six year period. Similar results were shown for both

males and females who were 9 and 12 years old at baseline. These results illustrate that

the probability of being active in adulthood is much higher if physical activity is

41

continuous during adolescence 69. A separate study using the Cardiovascular Risk in
Young Finns Study tracked physical activity behaviors over a 9-year period (1980-1989)
68. There was a 57% decline in the frequency of physical activity among males and 28%
decline among females over the 9 year period. Furthermore, the relative decline in
moderate and vigorous activity (frequency and intensity of physical activity combined)
was 55% in males and 20% in females 68.

Matton and colleagues 46 assessed the tracking of physical activity from
adolescence to adulthood in 138 females involved in the Leuven Longitudinal Study on
Lifestyle, Fitness, and Health. The physical activity behaviors of 14-18 year old
adolescent females were assessed via questionnaire that assessed sports participation,
physical activity at school, and sports participation during leisure time. In adulthood
(ages 37-43), physical activity was assessed by asking subjects to identify their three
most important sports. Subjects were then asked to report both frequency and duration of
each sport. During adolescence, girls who reported less than three hours of sports per
week were considered inactive, while those who reported three or more hours of work
were considered active. During adulthood, women reporting less than 1.5 hours of sports
per week were considered inactive while those participating in 1.5 or more hours per
week were considered active. During adolescence, 59% of girls were labeled active. Of
the girls who were labeled active, only 54% remained active in adulthood. Of the 41% of
adolescent girls who were labeled inactive, 63% remained inactive in adulthood. These
results suggest that physical activity levels do not remain stable between adolescence and

adulthood.

42

The National Longitudinal Study of Adolescent Health cohort is a nationally
representative school based study of more than 14,000 diverse youth between grades 7
and 12 which assesses the longitudinal trends in a variety of health behaviors of these
youth into young adulthood. Physical inactivity was assessed via questionnaire and was
defined as those adolescents who reported engaging in no weekly physical activity.
During adolescence, the prevalence of no weekly physical activity was 8% or less in each
female racial group. During young adulthood, the prevalence of physical inactivity
among different female racial groups increased between 31%-47%. For male
adolescents, the prevalence of reporting no weekly physical activity was 5% or less in
each racial group. This increased between 26%-3 0% during young adulthood. These
findings suggest a decline in physical activity fi'om adolescence to young adulthood on
ethnically diverse males and females 29.

In a separate study using the National Longitudinal Study of Adolescent Health
cohort, cluster analysis was used to identify meaningful patterns of physical activity
during adolescence and how these patterns are associated with healthy activity levels
during the transition to adulthood 49. Results identified seven different clusters of
adolescents: 1) adolescents with a high frequency of television and video games, 2)
adolescents who skateboard and play video games, 3) adolescents who play sports with
their parents, 4) adolescents who use neighborhood recreation centers, 5) adolescents
with parents who limit their television use, 6) adolescents who have control over their
television use, but watch very little, and 7) adolescents with high participation in school
activities. When compared to the television and video game group, the odds of meeting

physical activity recommendations (five or more weekly bouts of moderate to vigorous

43

physical activity) were highest among the skating/skateboarding group (OR = 13.1), the
play sports with parents group (OR = 5.8), the neighborhood recreation group (OR = 4.2),
and the high participation in school activities group (OR = 4.3). During young
adulthood, the odds of meeting physical activity recommendations declined in each
cluster, but remained significantly higher in the skating group (OR = 1.77), the play
sports with parents group (OR = 2.58), the neighborhood recreation group (OR = 2.26),
and the high participation in school activities group (OR = 2.35). These results illustrate
that within clusters of adolescents, physical activity patterns decline differently during the
transition to adulthood 49.

Tracking of physical activity between adolescence and young adulthood has also
been studied using data from the Northern Ireland Young Hearts Program. During
adolescence, daily participation 11 physical activity during a typical school day was
assessed via self-report questionnaire. During young adulthood, habitual work activity,
sports activity, and non-sports leisure activity were assessed 8’ 9. To assess the changes in
physical activity from adolescence to young adulthood, Boreham and colleagues 8
collected data from 245 males and 231 females at age 15 and during young adulthood.
Tracking of physical activity was assessed by dividing the subjects into three physical
activity classes: the lowest 25%, the middle 50%, and the highest 25%. Tracking of these
variables was assessed by determining the extent to which membership in a particular
class was maintained into adulthood using a kappa value. In males, the tracking of
physical activity was slight to fair (kappa = .202), but in females tracking was poor
(kappa = .021). These results illustrate there is a decline in physical activity patterns

during the transition form adolescence to young adulthood. In a separate study using the

44

Northern Ireland Young Hearts Project 9, the relationship between cardiovascular risk
factor status during young adulthood and adolescent physical activity was compared.
Although results showed no apparent relationships between physical activity during
adolescence and risk status during young adulthood, increased fitness levels during

adolescence were modestly associated with serum lipid levels and body fatness in young

adulthood.

Conclusion

There are few studies assessing change in physical activity behaviors among high
school and college students. In fact, the majority of research focuses on the general
adolescent and young adult populations. These studies consistently show a decline in
physical activity with age, but fail to target changes in physical activity levels that occur
specifically in the high school and college populations. If a more detailed understanding
of these changes can be established, it is possible to delay or prevent the decrease in
physical activity that occurs during this time period. This may lead to increase the

overall health and well-being of a college campus.

2.6 Changes in Physical Activity During College

The decline in physical activity is well documented throughout childhood and
adulthood 60. However, previous research illustrating what determines this decline is
limited as it may be specific to certain periods of life. Therefore, it is important to relate
the determinants of physical activity with certain age groups or focus specifically on

distinct developmental transitions during a lifespan. Healthy Campus 2010, which

45

includes physical activity as a leading health indicator, was created by to establish
national health objectives specifically for college students. The Healthy Campus 2010
document suggests that college students are an important population to study because it is
a specific time period in which an individual’s life is in transition 5. Thus, it is important
to focus on the determinants of physical activity during the time period when an
individual is enrolled in college.

Evidence suggests physical activity levels during the latter years of college may
continue into adulthood. Sparling and Snow 65 showed that 85% of college graduates
who were regular exercisers during their senior year of college were either as active or
more active five to ten years following graduation. Although these results suggest that
physical activity levels during the senior year of college may influence future physical
activity levels, it fails to assess the changes in physical activity participation over the
entire 4 years of enrolhnent in college.

Pinto and colleagues 57 assessed longitudinal changes in physical activity during
the beginning of college. Two hundred forty-two first-year college students enrolled in a
private university completed a baseline survey during their first year of college. A
follow-up survey was completed during the second year. Each study participant was
asked to report the total number of minutes spent in either: 1) vigorous physical activity
(exercise that made you sweat and breath hard) or 2) moderate physical activity (exercise
that did not make you sweat and breath hard). At baseline, average nrinutes spent in
vigorous activity (128 minutes) and moderate activity (97 minutes) were not significantly
different than the follow-up values for vigorous (176 minutes) and moderate (104

minutes) activity 57. These results suggest longitudinal changes in physical activity

46

during college are minimal. However, the results are limited to the time period between
the first and second year of college. Therefore, it is necessary to assess physical activity
throughout the entire 4-year period of college to have a more global perspective on
physical activity participation in college students.

To the best of our knowledge, no studies have assessed the changes in physical
activity and the corresponding determinants that result in these changes over four years of
enrolhnent in college. Thus, our understanding of the influence of college enrollment on
physical activity levels is limited and there is a need to develop research protocols to
assess this relationship. The purpose of the third aim of this project is to develop a
longitudinal research protocol which will track physical activity behavior and its
determinants effectively over a four year college career. Due to a lack of longitudinal
research on physical activity in college students, it is important to conduct new and
innovative research in this area to obtain a clear assessment of changes that occur in
physical activity during this period. If future research can better assess physical activity
behaviors throughout enrolhnent in college, universities will be able to more effectively
target those at risk for becoming physically inactive during college. As a result, physical
activity behaviors among college students will improve and these behaviors may continue

into adulthood.

47

CHAPTER 3
METHODS

3.1 Research Design

In brief, the research for this dissertation started with an undergraduate student
population at a large Midwestern University. Recruitment from a healthy lifestyles class
student population and a communications class student population made it possible to
compare these students cross-sectionally and after completion of their classes.
Assessment of these students included an initial cross-sectional survey followed by a re-
assessment at the end of the semester 15 weeks later and included analysis of cross-
sectional (baseline), as well as analysis of the longitudinal data. The first cohort included
students enrolled in a healthy lifestyles class recruited in September 2005 and followed
through December 2005. The second cohort included students enrolled in a healthy
lifestyles class recruited in January 2006 and followed through April 2006. The third
cohort included students enrolled in a communications class recruited in September 2005
and followed through December 2005. While these students were not enrolled at random
to their classes, one of the focal points of this research was to assess the impact of
enrollment in a healthy lifestyles class on physical activity energy expenditure compared
to students enrolled in a communications class. This research protocol had two major
elements: 1) completion of an anonymous online survey with standardized survey
questions and 2) a repetition of the online survey 15 weeks later. Figure 1 represents the

timeline of this research protocol.

48

Figure 1. Timeline for data collection

 

 

 

 

 

 

 

Cohort l Cohort 1 Cohort 2 Cohort 2
Healthy Lifestyles: Healthy Lifestyles: Healthy Lifestyles: Healthy Lifestyles:
Baseline Survey F ollow-up Survey F ollow-up Survey F ollow-up Survey
September December April 2006
2005 2005
Cohort 3 Cohort 3
Communications: Communications:
Baseline Survey Follow-up Survey
3.2 Study Population

The goal of this project was to determine if physical activity participation in
college students was improved via classroom based educational interactions that occur
during the college years. In addition, this project was designed to assess the impact of
gender, class membership, and high school sport and physical activity participation on
physical activity during college. Undergraduates enrolled in the healthy lifestyles class
(KIN 121) were designated as the target population of primary interest because these
students received educational interactions intended to influence their physical activity
levels. All kinesiology majors are required to take KIN 121, however non-kinesiology
majors are allowed to take the class as an elective. For a contrast with the experiences
received during enrollment in the healthy lifestyles class, a second target population was
identified as undergraduate students enrolled in a communications class (COM 225),
which did not include the activity and health component of the healthy lifestyles class.
All communication majors are required to take COM 225, however non-communication
majors are allowed to take the class as an elective. For the first cohort (KIN 12], fall

2005), the total number of students enrolled was 286. For the second cohort (KIN 121,

49

winter 2006), the total number of students enrolled was 305. For the third cohort (COM
225, fall 2005), the total number of students enrolled was 590. The total number of

students eligible to participate was 1240.

3.3 Eligibility Criteria (inclusion/exclusion)

All students enrolled in the healthy lifestyles or communications classes were
eligible for participation in this research. There were no exclusion criteria applied to these
undergraduate enrollees, unless a student was enrolled in both healthy lifestyles and
communications classes concurrently. Four individuals were enrolled in both classes

concurrently and their results were not included in the analysis.

3.4 Study Participants

Among the 1240 eligible to participate, a total of 911 (74%) completed the
baseline assessment and 765 (62%) completed the follow-up assessment. By cohort,
these numbers and proportions for baseline and follow-up were as followed: Cohort I
(KIN 121, fall 2005): 242 of 286 (85%), 193 of 286 (67%); Cohort 11 (KIN 121, winter
2006): 213 of 305 (70%), 177 of 305 (58%); Cohort III (COM 225, fall 2005): 456 of 590

(77%), 395 of 590 (67%).

3.5 Recruitment Procedures
Due to the fact that all data collection was anonymous, this study was granted
exempt approval status by the biomedical institutional review board at Michigan State

University. In brief, the steps of the recruitment process were as followed.

50

1)

2)

3)

4)

5)

6)

7)

8)

Consent forms (Appendix A) were distributed to all eligible study participants
enrolled in study during the first week of class.

The consent form included a survey admission ticket and instructions needed to
access the on-line survey which utilized the Longitudinal Surveillance Engine
(LSE). See 3.7 for a detailed description of the LSE assessment protocol.

Verbal instructions were given by a study investigator to help ensure
understanding of how to access the online survey.

Students agreeing to participate used their unique LSE admission tickets to
complete the online assessment.

Study participants were given 10 days to complete the baseline questionnaire. To
encourage participation, a reminder e-mail was sent to all study participants on
the 5th, 7‘“, and 9th days of the 10 day period.

A second series of reminder e-mails were sent during the 15th week of the study
period encouraging subjects to participate in the follow-up survey.

Study participants were given 10 days to complete the follow-up questionnaire. A
reminder e-mail was sent to all study participants on the 5th, 7‘“, and 9th days of
the 10 day period.

Upon completion of the baseline and follow-up surveys, students were eligible to
receive a coupon for a free ice cream cone at the Dairy Store located at Michigan

State University.

51

3.6 Over-Arching Conceptual Model

The basic conceptual model for Study Aim 1 and 2 are depicted in Figures 2 and
3, respectively. At the right of the figures are the primary outcome variables of interest
(e. g. college physical activity participation). At the left of the Figure 2 are the exposure
variables that may influence the physical activity outcome variable. The primary
exposure variables that were addressed in sub-aim 1 (gender, class) are shown at the top
left of Figure 2. The primary exposure variables that were addressed in sub-aim 2 (high
school sport participation, high school leisure time physical activity) are shown at the
bottom left of Figure 2. All additional exposure variables shown in Figure 2 were
included in the analyses. The arrows at the far left represent the interactions that were
assessed. At the far left of Figure 3 are sub-sets of time invariant covariates (e.g. gender,
class, etc.) that could influence these response variables either directly or indirectly. The
time invariant covariates reflect exposure variables that do not change over time.
Midway between the time-invariant covariates and the response variables are tirne-variant
covariates of interest. The time-variant covariates can vary over time and it is possible
that changes in these variables can influence change in physical activity participation.

The arrows at the far left of Figure 3 represent the interactions that were assessed.

52

Figure 2: Over-arching conceptual model for Study Aim 1

 

 

Exposure Variables Outcome variable

 

 

 

 

 

Sub-aim 1 Primary
Exposure Variables

Gender

 

 

 

 

Class

 

 

 

Additional Exposure
Variables

 

 

 

Age

 

Race

 

Major

 

 

Physical activity

Lrvmg status participation

 

 

 

 

BMI

 

Television
viewing

 

[\

Video game
use

 

 

 

 

 

Sub-aim 2 Primary
Exposure Variables

High school /
sport

participation

 

 

 

High school
LTPA
participation

 

 

 

53

Figure 3: Over-arching conceptual model for Study Aim 2

 

Time invariant Time variant Primary outcome
covariates covariates variables

 

 

 

 

 

Primary Exposure
Variable of Interest

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

    

 

 

 

 

 

  
 

 

 

 

Class

BMI \

. . Change in

Televrsron physical activity
Gender vrewrng participation
Age
Race ll‘k Video game

[[117 use
Major ll,
. , j .l

Lrvrng status 7
High school
physical
activity and
sport
participation

 

 

 

3.7 Assessment Protocol:

This research project utilized the Longitudinal Survey Engine (LSE) which was
recently developed at Michigan State University. The LSE is an intemet based survey
engine that offers anonymity and privacy for exploring topics that might be too sensitive
to effectively engage participants in scientifically meaningful disclosures. The LSE is

designed for large sample longitudinal, research protocols. Under the LSE protocols,

54

participants log in anonymously to create their own userID and passwords, and complete

an initial baseline survey. Thereafter, participants are invited to return for follow-up

surveys using the previously created userID and passwords. The userID/password feature

allows the investigator to monitor individual and group patterns of health and behaviors

anonymously over time. The LSE can be menu-driven using major intemet browsers

(Internet Explorer, Netscape, etc.). The main features of the LSE used in the current

investigation are:

a)

b)

Each study participant was provided online access to the LSE
(http://lse.commtechlab.msu.edu/login.php)

The LSE software used a unique survey ID number to track each individual survey
protocol (e. g., an initial online baseline questionnaire or assessment followed by
longitudinal assessments), and to track each participant's responses at baseline and
over time. (The LSE built a unique ID number to track both the survey protocol and
the individual participant. This ID number was called a 'coupon number.')

Prior to completion of the initial baseline assessment, the designated participant
accessed the LSE website, keyed in his or her LSE coupon number, and was
prompted to create his or her own userID and password. This was followed by
prompting questions to help the individual recall the userID and password if these
were lost or forgotten (e. g., 'What is the first initial of your mother's maiden name
(surname before marriage)?’ This approach was used so designated participants could
log in anonymously without direct or indirect linkage between the participant's

identity and the survey responses.

55

d) Thereafter, the web browser displayed a study disclosure statement which invited
participation. The participant could key the NEXT button to proceed to complete the
survey, or could log out at any time. The LSE was designed to allow the participant to
terminate participation and/or to skip over a survey item.

e) Survey elements used in this investigation included true/false, multiple choice, and
short answer items.

f) The LSE kept track of successive log-ins by designated participants, which allowed
them to complete repeated measurements. Storage of the repeated measures occurred
within a database that encoded the userlD and a date/time stamp. Number of logins
and the interval between repeated measures were controlled by the investigator via
options given by the LSE. All responses were stored within a protocol-specific

EXCEL Spreadsheet on the LSE secure server.

3.8 Primary Outcome Variables

College leisure time physical activity participation: At baseline, each study
participant was asked if s/he participated in any leisure-time physical activities or
exercises such as running, calisthenics, golf, gardening, walking, etc. during the past
month. If the study participant answered yes, a series of three remaining physical activity
questions were asked to assess the type, frequency (sessions per week), and the duration
(average minutes per session) of activity. These questions are based on similar

previously validated instruments " 3' 33

. If the study participant reported participating in a
second or third activity, the same series of questions were repeated for each activity. The

same series of questions were repeated at the end of the 15-week testing period. The

56

survey is shown in appendix B. To determine if expressing the outcome variable
differently would alter the results, college physical activity was expressed three ways.
First, physical activity was expressed as a continuous variable calculated as kilocalories
expended per week indexed by body weight (kcal/kg/week). A metabolic equivalent
value (MET) was assigned for each activity using the Compendium of Physical Activities
2. One MET is equivalent to one kcal/kg/hour. Therefore, using the MET values for
each activity, the number of sessions per week for the activity, and the number of minutes
per session, the total number of METS per hour were determined and extrapolated into
weekly caloric expenditure (kcal/kg/week). Second, physical activity was expressed as
an ordinal variable calculated as quartiles of kcal/kg/week (Quartile 1: 0-12.4
kcal/kg/week, Quartile 2: 12.5-25.4 kcal/kg/week, Quartile 3: 255-458 kcal/kg/week,
Quartile 4: 45.9 or more kcal/kg/week). Third, physical activity was expressed as a three
level categorical variable based on the ACSM recommendations for moderate and
vigorous physical activity 4 (Inactive: reported no leisure time activity or did not meet the
physical activity recommendations, Moderately Active: engaged in moderate physical
activity for at least 30 minutes on five or more days per week , Vigorously Active:
engaged in vigorous physical activity three or more days per week for 20 or more
minutes per occasion).

Change in college physical activity participation: In order to assess change in
caloric expenditure during the 15-week study period, the difference between physical
activity levels at baseline and follow-up were assessed (Study aim 2). To determine if
expressing the outcome variable differently would alter the results, the change in physical

activity was expressed three ways: 1) as a continuous variable (difference in

57

kcal/kg/week at the beginning of the semester and kcal/kg/week at the end of the
semester), 2) as a categorical variable using the physical activity quartiles (Decline: study
participants who moved to a lower physical activity quartile at the end of the semester,
Stay the same: study participants who remained in the same quartile throughout the
semester period, Increase: study participants who moved into a higher physical activity
quartile), 3) as a categorical variable using physical activity recommendations (Decline:
study participants who remained in the inactive category, decreased fiom vigorously
active to inactive, or decreased from moderately active to inactive, Stay the same: study
participants who remained in either the inactive, moderate, or vigorous category, or
moved from the moderate to vigorous category or vigorous to moderate category,
Increase: study participants who increased from the inactive category to either the

moderate or vigorous category).

3.9 Definitions for Exposure Variables

Gender: Gender was defined as a dichotomous variable (male/female).

Class membership: Class membership was defined as a dichotomous variable
(healthy lifestyles/communications).

Age: Age was defined as a categorical variable (18 years, 19 years, 20 years, 21
years, 22 or more years)

Race: Race was defined as a categorical variable (Caucasian, African American,

Other (e.g., Hispanic, Asian, Pacific Islander, Native American)).

58

Living Status: Current living status was defined as a dichotomous variable (living
on-campus, living off-campus). Data on living status were collected at the beginning and
end of the semester.

Living Status change: Change in living status during the study period was
assessed as a categorical variable: students whose living status did not change (no
change), students whose living status changed from living on-campus to living off-
campus (on/off-campus), and students whose living status changed from living off-
carnpus to living on-campus (off/on-campus).

Major: Major was defined as a dichotomous variable (kinesiology/other).

Body Mass Index: Data on self-reported weight 0g) and height (m) were
collected and body mass index (kg/m2) was calculated. For analysis, BMI was specified
as a categorical variable based on the Expert Panel on the Identification, Evaluation, and
Treatment of Overweight and Obesity on Adults 52. Subjects with a BMI between 18.5
and 24.9 were considered normal weight, between 25 and 29.9 were overweight, and 30
or greater were considered obese. Less than 2% of the population had a BMI less than
18.5 and were included in the normal weight group. Body mass index data were
collected at the beginning and end of the lS-week study period.

BMI change: Change in BMI during the study period was calculated by taking the
difference of BMI at the beginning and end of the semester (BMI at the end of the
semester — BMI at the beginning semester). A three level categorical variable was used
to describe change in BMI: increase (a increase in BMI during the semester), stay the
same (BMI that did not change during the semester), decrease (a decrease in BM] during

the semester).

59

Television viewing: Television viewing was categorized as an ordinal variable
(less than 2 hours per day, 2 hours per day, 3 hours per day, 4 hours per day). Television
viewing data were collected at the beginning and end of the 15-week study period.

Television viewing change: The change in television viewing during the study
period was calculated by taking the difference of the amount of television watched at the
beginning and end of the semester (TV hours at the end of the semester - TV hours at the
beginning semester). A three level categorical variable was used to describe change in
TV hours: increase (a increase in TV hours during the semester), stay the same (TV hours
that did not change during the semester), decrease (a decrease in TV hours during the
semester).

Video game use: Video game use was categorized as an ordinal variable (0 hours
per day, 1 hour per day, 2 or more hours per day). Data on video game use was collected
at the beginning and end of the 15-week study period.

Video game use change: The change in video game use during the study period
was calculated by taking the difference of the amount of video game use at the beginning
and end of the semester (VG hours at the end of the semester — VG hours at the beginning
semester). A three level categorical variable was used to describe change in VG hours:
increase (a increase in VG hours during the semester), stay the same (VG hours that did
not change during the semester), decrease (a decrease in VG hours during the semester).

High school sports: Participation in school sponsored sport activity was assessed
by asking each subject if s/he participated in any school sponsored and/or club Sports or
activities (e. g. basketball, cheerleading, marching band, etc.) during high school. If the

student participated in a school sponsored sport, s/he was first asked to identify each high

60

school sport in which s/he participated. This was followed by asking: 1) how many
months per year, 2) the average number of sessions per week, and 3) the average minutes
per session s/he participated in these school-sponsored sports combined. Using MET
values for each physical activity, average weekly caloric expenditure (kcal/kg/week) for a
sponsored high school sports was calculated. In this study, high school sport
participation was defined as the number of high school sports played (0 sports, 1 sport, 2
sports, 3 sports, 4 or more sports).

High school leisure time physical activity participation flrigh school LTPA):
Participation in high school leisure time physical activity was assessed by asking if each
subject participated in any non-school sponsored leisure-time physical activities or
exercises (any activity that was not sponsored directly by the high school) such as
nmning, recreational sports, gardening, etc. during high school. This was also followed
by a series of questions asking: 1) months per year, 2) average number of sessions per
week and 3) average minutes per session s/he participated in these non school-sponsored
physical activities combined. Using MET values for each physical activity, study
participants were categorized as: 1) sedentary (reported no non-school sponsored sport
activity, 2) inactive (reported some leisure time activity but did not meet the
recommendations for moderate or vigorous activity), 3) moderately active (engaged in
moderate physical activity (3-6 METS) for at least 30 minutes on five or more days per
week) or 4) vigorously active (engaged in vigorous physical activity (greater then 6

METS)three or more days per week for 20 or more nrinutes per occasion).

61

3.10 Study Aim 1

The overall purpose of Study Aim 1 was to examine leisure time physical activity
participation in college students. This aim had two sub-aims. Sub-aim 1 described
college students’ current physical activity levels in relation to class membership (healthy
lifestyles class/communications class) and gender. Sub-aim 2 assessed impact of
participation in high school sponsored sports and non-high school sponsored sports and
activities on physical activity participation during college. An underlying objective of
Study Aim 1 was to determine if using different physical activity outcome variables
would alter the results of our analyses. Therefore, physical activity was assessed in three
different ways: 1) caloric expenditure (i.e., continuous), 2) quartiles for physical caloric
expenditure (i.e., ordinal), and 3) meeting national recommendations for vigorous or

moderate physical activity (i.e., categorical).

3.11 Statistical Analysis for Study Aim 1:

In order to assess the impact of class membership on college physical activity
participation, statistical power was determined assuming that data would be available
from 450 students enrolled in the healthy lifestyles and 450 students enrolled in the
communications classes, leisure time physical activity (LTPA) specified as kcal/kg/week
was the outcome variable, and a 20% difference in physical activity is clinically
meaningful. Using BRFSS data from Michigan used in a previous study (personal
communication, Pivamik), LTPA energy expenditure averaged 24.0 kcal/kg/wk in 18-22
year-old respondents with a standard deviation of 25.4. Therefore, the following

assumptions were used to calculate statistical power: 1) average LTPA for the healthy

62

lifestyles class is 25.0125 kcal/kg/wk, 2) average LTPA for the communications class is
20.0:t25 kcal/kg/wk, and 3) a two-tailed test with an alpha level of 5%. Statistical power
for detecting a difference of 5 kcal/kg/wk is 0.85.

All analyses were completed using SPSS 14.0. The first step in the analysis was
to ensure a high quality dataset free of miscoded or erroneously missing values.
Frequency distributions were assessed which disclosed the need for any recoding or
transformations in advance of statistical analysis. After examination of the distribution of
the continuous physical activity outcome variable (kcal/kg/week), it was determined that
the variable was positively skewed and a transformation was needed. The log
transformation of the kcal/kg/week variable normalized the data and was used in all
analysis. The results of this transformation are shown in Appendix C.

Initial descriptive analysis of the continuous outcome variable included one way-
AN OVA. This approach was used to assess statistical differences (p<.05) in leisure time
physical activity caloric expenditure (log of kcal/kg/min) for each categorical exposure
variable (gender age, race, living status, major, body mass index, television, viewing, and
video game use) at baseline. A linear regression model was used to assess the impact of
class membership and gender on the log of kcal/kg/min after adjustment for age, race,
living status, major, body mass index, television, viewing, video game use, and high
school physical activity and sport participation. All exposure variables were entered into
the model regardless of p-value in a backwards stepwise fashion. The variables were
removed one at a time based on the largest p-value. Any exposure variable with a p-
value <.05 was included in the final model. The following interactions were assessed in

this and all subsequent analyses for Study Aim 1: gender/class, gender/age, gender/high

63

school sport, gender/high school LTPA, class/age, class/major, class/race, high school
sport/high school LTPA, and television viewing/video game use.

Ordinal logistic regression is used for analysis of data on an ordinal scale as it
captures the ranked nature of the dependent variable. In this analysis, quartiles of
physical activity is the ordinal outcome variable of interest. Prior to the regression
analysis, cross-tabulations were generated and the chi square test for linear trend was
used to determine any significant association (p<.05) between physical activity quartile
and the exposure variables of interest. Using ordinal logistic regression, unadjusted odds
ratios were calculated to determine the impact of each covariate on membership to a
higher physical activity quartile. That is, this odds ratio applies to any adjacent quartile
(i.e., Q1 compared to Q2-4, Q1-2 compared to Q3-4, or Q1-3 compared to Q4) and is the
common odds ratio across each dichotomization. A positive odds ratio indicates an
increased likelihood that a specific response to an exposure variable (e.g., male) will be
observed in a higher category. A negative coefficient indicates a decreased likelihood that
a Specific response to an exposure variable will be observed in a lower category. Using
the ordinal logistic regression model, adjusted odds ratios were calculated to determine
the impact of each covariate on membership to any higher physical activity quartile after
adjusting for covariates. Each covariate was entered into the model and order of entry
was dictated by the level of the p-value for the exposure variable. The underlying
assumption of ordinal logistic regression is that the coefficients or odds ratios that
describe the relationship between categories of the response variable are the same
(proportional odds assumption). As a result, only one model (one odds ratio) is needed to

describe each covariate in the model. In SPSS, the test of parallel lines is used to assess

64

the assumption of proportional odds. A p-value < .05 suggests that the model does not
meet the proportional odds assumption. In the event that the proportional odds
assumption was not met, the entire model was replaced and a multinomial logistic
regression was used to analyze the data for all variables.

Multinorrrial logistic regression was used to assess the three level categorical
variable for meeting the recommendations for physical activity (i.e., inactive, moderately
active, vigorously active). Cross-tabulations were calculated and chi square was used to
determine statistical associations between meeting the recommendations for physical
activity and the exposure variables of interest. Using multinomial logistic regression,
unadjusted odds ratios were calculated to determine the impact of each covariate on
meeting either the recommendations for moderate physical activity or vigorous physical
activity. The inactive category served as the referent group. Adjusted odds ratios were
then generated using multinomial logistic regression. Due to the small number of
exposure variables, all were entered into the model. Gender, age, and race, were forced
into the model while class, major, living status, television viewing, video game use, BMI,
high school sponsored sport participation, and non-high school sponsored LTPA were
entered regardless of level of significance in a backwards stepwise fashion. The variables
not forced into the model were removed one at a time based on the largest p-value. Any
exposure variable with a p-value <.05 was included in the final model. The interactions

between the exposure variables previously described were assessed.

65

3.12 Study Aim 2

The overall purpose of Study Aim 2 was to assess the impact of college students’
enrollment in a semester long healthy lifestyles class on change in physical activity
participation. Physical activity participation was assessed three ways: 1) change in
caloric expenditure (kcal/kg/week) during the semester, 2) change in quartile of physical
caloric expenditure during the semester, and 3) change in meeting the national

recommendations for physical activity during the semester.

3.13 Statistical Analysis for Study Aim 2:

In order to assess the impact of college students’ enrollment in a semester long
healthy lifestyles class on physical activity participation, statistical power was determined
assuming that: 1) complete data (two surveys over three months) would be available fiorn
250 students, and that 2) leisure time physical activity (LTPA) specified as kcal/kg/week
is the primary endpoint, and 3) a 20% increase is clinically meaningful. Using BRF SS
data from Michigan in a previous study (personal communication, Pivarnik), LTPA
energy expenditure averaged 24.0 kcal/kg/wk in 18-22 year-old respondents with a
standard deviation of 25.4. Thus, the statistical power for detecting an increase of 5
kcaI/kg/wk (two-tailed test with an alpha level of 5%) was 0.88. All analyses were
completed using SPSS 14.0. The same procedures shown in section 3.11 were used to: 1)
ensure a high quality dataset and 2) assess the frequency distributions. The change in log
kcal/kg/week was used in analysis of the continuous variable.

One way-ANOVA was used to assess significant differences (p<.05) in the

change in caloric expenditure over a semester, calculated as the difference between

66

beginning and end of semester kcal/kg/week (follow-up kcal/kg/week — baseline
kcal/kg/week) , for class membership, gender, age, race, living status, major, body mass
index, television, viewing, and video game use. A multiple linear regression model was
used to assess the impact of class membership on change in log transformed caloric
expenditure when specified as a continuous variable. The time invariant covariates class,
gender, age, race, living status, and major, and time variant covariate body mass index,
television, viewing, and video game use were entered into the model in a backwards
stepwise fashion. All exposure variables were entered into the model regardless of p-
value in a backwards stepwise fashion. The variables were removed one at a time based
on the largest p-value. Any exposure variable with a p-value <.05 was included in the
final model. The time invariant and variant covariates were not handled differently in the
model. The following interactions were assessed in this and all subsequent analyses for
Study Aim 2: gender/class, gender/age, class/age, class/major, class/race, and television
viewing/video game use.

Multinorrrial logistic regression was also used to assess the change in physical
activity quartile and change in physical activity category during the semester. Cross-
tabulations were calculated and the chi square was used to determine any significant
association (p<.05) between these two variables and the exposure variables/covariates.
Using multinomial logistic regression, unadjusted, bivariate odds ratios were calculated
to assess the impact of each covariate on: 1) moving to a lower physical activity quartile
or category, or 2) moving to a higher physical activity quartile or category during the
semester. Those individuals who remained in the same physical activity quartile or

category during the semester served as the referent group. Adjusted odds ratios were

67

then calculated to determine the impact of each covariate on change in physical activity
quartile after adjustment for covariates. Gender, age, and race, were forced into the
model while class, major, living status, television viewing, video game use, and BMI,
were entered into the model in a backwards stepwise fashion regardless of level of
significance. The variables not forced into the model were removed one at a time based
on the largest p-value. Any exposure variable with a p-value <.05 was included in the
final model. The interactions between the exposure variables previously described were

assessed.

3.14 Supplementary Statistical Analysis

A series of supplement analyses were performed to address the limitations of the
statistical analyses previously described. The limitations are described in detail within
this section and the results of the additional statistical analyses are reported in Chapter 4.

The inability to account for the attrition rate in subject participation from the
baseline to follow-up surveys was the first limitation in the prior modeling strategies.
The previous analysis used a change score, which was calculated as the difference in
physical activity energy expenditure between baseline and follow-up. Therefore, the
analysis was restricted to those students who completed both the baseline and follow-up
surveys and data were excluded from 146 subjects. An alternative to this method is to
assess the impact of the enrollment in a healthy lifestyles class using a multivariate model
which allows for an outcome response variable that contains both baseline and follow-up
physical activity values. Although this analysis does not formally deal with the attrition

between baseline and follow-up, it uses all available data at baseline to assess the contrast

68

in physical activity energy expenditure between students enrolled in the healthy lifestyles
class and the communications class. This analysis allows us to test if the contrast in
energy expenditure between the two classes at follow-up (which includes data only fiom
students who completed the follow-up survey) is greater than the contrast at baseline
(which includes data from all students who completed the baseline survey). There are
two alternatives when conducting this analysis. The first alternative is to estimate the
population-averaged estimates of the effect of the healthy lifestyles class on physical
activity levels. Generalized linear models and generalized estimating equations were
used to estimate this effect. This estimate formally tests if the magnitude of difference in
physical activity between the healthy lifestyles and communications at the end of the
semester is larger than the size of the difference between these two classes at baseline.
The population-based approach estimates the effect of taking the entire sample and
assigning them to either the healthy lifestyles class or the communications class. The
second alternative is to estimate the subject-specific effect of the healthy lifestyles class
on levels of physical activity. A random intercept version of the generalized linear model
was used to estimate this effect. This method estimates how much improvement in
physical activity a college student can receive if s/he were assigned to the healthy
lifestyles class rather than the communications class. Stata 8.0 was used for all analyses.
The value of the calculated change score was the second limitation that needed to
be addressed. Change scores do not take into account imbalances at baseline between the
students enrolled in the healthy lifestyles and communications classes. Because of
regression to the mean, it is possible that students with lower baseline physical activity

energy expenditure values are more likely to improve than those with higher values. An

69

alternative to the change score approach is to use the follow-up energy expenditure value
as the outcome variable and assess the impact of the exposure variables such as gender
and class while holding the baseline energy expenditure value constant. This analysis
was completed using a general linear model.

The third limitation to be addressed was the nested structure of the healthy
lifestyles class. That is, there were Six different sections of healthy lifestyles classes
during each semester. It is possible that students enrolled in a given section of the
healthy lifestyles class are more alike than students enrolled in a different section of the
healthy lifestyles class. As a result, it is important to determine if physical activity
energy expenditure during enrollment in the healthy lifestyles class is unique to specific
sections of the healthy lifestyles class. In order to address this limitation, a generalized
linear model was used and students were clustered by healthy lifestyles section to
determine if results changed when controlling for the nested structure of this class.

A fourth limitation addressed was the use of the class membership variable that
did not distinguish between students enrolled in the healthy lifestyles class during the fall
semester and those enrolled in the healthy lifestyles class in the spring semester. It is
possible that change in physical activity participation during the semester is different
between students enrolled in these two classes. Generalized linear models and
generalized estimating equations were conducted using a three level categorical variable
for class membership that distinguished between students enrolled in the fall and spring

healthy lifestyles classes.

70

CHAPTER 4
RESULTS

4.1 Descriptive Data

Among the 1240 students eligible to participate, a total of 911 (74%) completed
the baseline assessment and 765 (62%) completed the follow-up assessment. By cohort,
these numbers and proportions were as follows: Cohort I (KIN 121, fall 2005): 242 of
286 (85%); Cohort 11 (KIN 121, winter 2006): 213 of 305 (70%); Cohort [11 (COM 225,
fall 2005): 456 of 590 (77%). Due to the similarities between the two Healthy lifestyles
cohorts, they were combined for all subsequent analyses. Descriptive data for the Fall
2005 and Spring 2006 healthy lifestyles classes are shown in Table 1 to illustrate these
similarities. There were no statistically significant differences in physical activity,
gender, race, living status, BMI, and high school physical activity participation. There
were significant differences in age and major as the Spring 2006 healthy lifestyles class
enrolled a lower proportion of 18 year olds and kinesiology majors. The change in
physical activity energy expenditure (change in log kcal/kg/week) approached
statistically significant differences between the two healthy lifestyles classes. On
average, students enrolled in the fall semester decreased energy expenditure during the
semester whereas the spring semester students did not. This difference is discussed in the

supplementary analysis section (Section 4.14).

71

Table 1. Descriptive data comparison for the healthy lifestyles classes

 

 

Fall 2005 Sprmg3006
n % n % Chi-square p-value
Age
22 years or more 36 14.9 29 13.6 0.002
21 years 43 17.8 45 21.1
20 years 48 19.8 46 21.6
19 years 42 '1 7.4 60 28.2
18 years 73 30.2 33 15.5
Gender
Male 86 35.5 61 28.6 0.132
Female 156 64.5 152 71.4
Race
Afiican American 22 9.1 25 11.7 0.286
Other 26 10.7 15 7.0
Caucasian 194 80.2 173 81.2
Major
Kinesiology 121 50.4 68 32.2 <.001
Other 1 19 49.6 143 67.8
Living Status
On-campus 122 5 1 .3 100 47.2 0.397
Off-campus l 16 48.7 1 12 52.8
BMl
Obese 13 5.5 9 4.3 0.744
Overweight 53 22.5 52 24.6
Normal weight 170 72.0 150 71.1
High school sports
4 or more 36 14.9 30 14.1 0.616
3 sports 65 26.9 53 24.9
2 sports 60 24.8 50 23.5
1 sport 39 16.1 47 22.1
0 sports 42 17.4 33 15.5
High school LTPA
Vigorous 84 34.7 65 30.7 0.377
Moderate 22 9. l 19 9.0
Inactive 62 25.6 70 27.4
Sedentary 74 30.6 58 46.7
Physical activity quartile
0-12.5 kcang/week 49 20.2 45 21.4 0.503
12.5 - 25.5 kcal/kg/wk 53 21.9 53 25.2
25.5 - 45.9 kcal/kg/wk 61 25.2 57 27.1
>45.9 kcal/kg/wk 79 32.6 55 26.2
Physical activity
recommendations
Inactive 35 14.5 38 18.0 0.268
Moderate 36 14.9 22 10.4
Vigorous 171 70.7 151 71.6
mean SD mean SD p-value
Log kcaI/kg/wk 3.40 0.91 3.25 0.88 0.273
Chflge in Iowawgwk -0.15 0.74 0.02 0.78 0.050

 

72

Descriptive data for age, gender, race, major, living status, and BMI category are
shown in Table 2. Chi square analysis was used to assess the statistical differences
between the healthy lifestyles and communications classes. The majority of the sample
(73%) was between the ages of 18 and 20 years, female (65%), Caucasian (84%), and
normal weight (72%). An even proportion of students lived on-campus and off-campus.
Nearly 41% of students enrolled in the healthy lifestyles class were kinesiology majors
compared to only 1% of the students enrolled in the communications class. After
formally testing if physical activity levels were different between the kinesiology majors
and non-kinesiology majors, our results showed that physical activity levels in
kinesiology majors (46. 1:8 7.5 ) were significantly higher (p<.05) than the non-majors

(31.6:|:28.2) enrolled in the healthy lifestyles class.

73

Table 2: Descriptive data for the study population

 

 

Healthy Chi-
Total Lifestyles Communications square
(n=9l 1) (n=45 5) (n=456) p-value
n % n % n %

Age
22 years or more 101 11.1 65 14.3 36 7.9 <.001
21 years 143 15.7 88 19.3 55 12.1
20 years 185 20.3 94 20.7 91 20.0
19 years 257 28.2 102 22.4 155 34.0
18 years 225 24.7 106 23.3 119 26.1

Gender
Male 321 35.2 147 32.3 174 38.2 0.038
Female 590 64.8 308 67.7 282 61.8

Race
Afiican American 71 7.8 47 10.3 24 5.3 0.014
Other 79 8.7 41 9.0 38 8.3
Caucasian 761 83.5 367 80.7 394 86.4

Major
Kinesiology 193 21.2 189 41.5 4 0.9 <.001
Other 714 78.4 262 57.6 452 99.1

Living Status
On-campus 439 48.2 222 48.8 217 47.6 0.323
Off-campus 467 51.3 228 50.1 239 52.4

BMI
Obese 50 5.5 22 4.8 28 6.1 0.052
Overweight 181 19.9 105 23.1 76 16.7

Normal weight 659 72.3 320 70.3 339 74.3

 

4.2 Description of Study Aim 1

The first Study Aim assessed the effect of gender, class membership, and high
school physical activity and sport participation on physical activity levels in college. An
underlying objective of Study Aim 1 was to assess if the impact of these exposure
variables change when using different physical activity outcome measures. Therefore,
Study Aim 1 utilized a continuous variable (log transformed kcal/kg/wk), ordinal variable

(quartiles of kcal/kg/wk), and categorical variable (meeting the recommendations for

74

moderate or vigorous physical activity) for all analyses. Specifically, Study Aim 1 for
this dissertation was:

Study Aim 1 (sub-aim 1): Does gender or class membership influence physical
activity energy expenditure in college students?

Study Aim 1 (sub-aim 2): Does high school sport and physical activity

participation influence physical activity energy expenditure in college students?

4.3 Assessment of Physical Activity Outcome Variable

It was determined that results remained essentially the same regardless of how
physical activity was specified. That is, the impact of the exposure variables on physical
activity participation during college was similar whether using the continuous,
categorical, or ordinal physical activity outcome variable. Therefore, we chose to use the
models based on the ordinal outcome, physical activity quartiles, due to its ease of
interpretation. Physical activity quartiles were created based on the physical activity
caloric expenditure results (kcal/kg/wk). Figure 4 Shows the fi'equency distribution and
resultant quartiles for caloric expenditure in college students at entry into the cohort.
Quartile 1 (Q1) corresponds to a caloric expenditure between 0 and 12.5 kcal/kg/wk,
quartile 2 (Q2) between 12.5 and 25.5 kcal/kg/wk, quartile 3 (Q3) between 25.5 and 45.9
kcal/kg/wk, and quartile 4 (Q4) 45.9 or more kcal/kg/wk. All subsequent analyses in this
project will use this ordinal physical activity outcome variable as well. The results for
the continuous physical activity outcome variable and the categorical physical activity

outcome variable are shown in Appendix D and Appendix B, respectively.

75

Figure 4: Graphical representation of physical activity quartile data

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

won or 02 03 Q4
801
3 60~ H
E '1
3
6'
O
L.
IL
40‘!
1
20d :—
0-—- -- ~ ~ ~. ~
0.00 50.00 1001!) 150.00

kcallkglwook

4.4 Study Aim 1 Descriptive Results

The chi square test for linear trend was used to assess the linear relationship
between each predictor variable and quartile of physical activity at the beginning of the
semester (Table 3). There are statistically significant linear relationships between
quartile of physical activity and gender, class membership, major, race, age, living status,
television viewing, video game use, number of high school sponsored sports played, and
participation in non-sponsored high school physical activity. These statistically

significant trends illustrate that physical activity was higher in males, students enrolled in

76

the healthy lifestyles class, kinesiology majors, Caucasians, and those who were more
active in high school sponsored sports and high school LTPA. Physical activity was

lower in older students and those who watched more television.

77

Table 3. Descriptive data for quartiles of plysical activity at the beginning of a semester
0-12.5 12.5 - 25.5 25.5-45.9 >45.9 p-value
kcal/kg/wk kcal/kg/wk kcal/kg/wk kcal/kg/wk for

 

 

Variable N N % N % N % N % trend
Gender
Male 321 43 13% 77 24% 76 24% 125 39% 0.000
Female 586 184 31% 152 26% 149 25% 101 17%
Class

Healthy Lifestyles 452 94 21% 106 23% 118 26% 134 30% 0.000
Communications 455 133 29% 123 27% 107 24% 92 20%

Major
Kinesiology 193 25 13% 42 22% 50 26% 76 39% 0.000
Other 710 202 28% 186 26% 1 73 24% 149 21%

Race
African American 71 38 54% 9 13% 7 10% 17 24% 0.000
Other 78 26 33% 23 29% 12 12% 17 22%
Caucasian 759 163 21% 197 26% 206 27% 192 25%

Age
22 orolder 101 29 29% 33 33% 21 21% 18 18% 0.003
21 141 39 28% 38 27% 31 22% 33 23%
20 185 44 24% 45 24% 53 29% 43 23%
1 9 25 5 67 26% 66 26% 62 24% 60 24%
18 225 48 21% 47 21% 58 26% 72 32%

Living Status
On-campus 437 106 24% 97 22% 110 25% 124 28% 0.035
Off-campus 465 121 26% 129 28% 1 14 25% 101 22%

Television viewing
4 or more hours 71 24 34% 17 24% 14 20% 16 23% 0.024
3 hours 118 36 31% 29 25% 23 19% 30 25%
2 hours 236 60 25% 62 26% 60 25% 53 22%
less than 2 hours 483 107 22% 121 25% 128 27% 127 26%
Video game use

0 hours 639 178 28% 162 25% 164 26% 135 21% 0.000
1 hour or less 208 31 15% 54 26% 51 25% 72 35%
2 or more hours 60 18 30% 13 22% 10 17% 19 32%

BM]
Obese 49 15 31% 1 l 22% 12 24% l 1 22% 0.810
Overweight l 80 46 26% 47 26% 30 1 7% 57 32%
Normal weight 657 162 25% 166 25% 181 28% 148 23%

High school sports
4 or more 1 17 20 17% 26 22% 27 23% 44 38% 0.000
3 sports 222 46 21% 52 23% 63 28% 61 27%
2 sports 234 52 22% 67 29% 61 26% 54 23%
1 Sport 186 55 30% 50 27% 40 22% 41 22%
0 sports 148 54 36% 34 23% 34 23% 26 18%

High school LTPA
Vigorous 279 39 14% 51 18% 85 30% 104 37% 0.000
Moderate 75 9 12% 16 2 1% 18 24% 32 43%
Inactive 276 78 28% 90 33% 68 25% 40 14%
Sedentag 276 1 01 3 7% 72 26% 53 19% 50 1 8%

 

78

4.5 Test of the Proportional Odds Assumption.

The underlying assumption of ordinal logistic regression is that the odds ratios
describing the relationship between categories of the response variable are the same
across quartiles of physical activity (proportional odds assumption). That is, the odds
ratios for the variables in the equations would not vary significantly if they were
estimated separately across different dichotomizations of the physical activity quartile
data. As a result, only one model (one odds ratio) is needed to describe the data. In
SPSS, the test of parallel lines is used to assess the assumption. A p-value <.05 indicates
that the model does not meet the proportional odds assumption and a different model
such as a multinomial logistic regression should be used to describe the data. Table 4
illustrates that gender, race, and BMI do not meet this assumption. As an example, a
graphical representation of the proportional odds assumption is shown for race and class
in Figures 5 and 6. In these analyses, quartile 1 was compared to quartiles 2-4, quartiles
1 and 2 were compared to quartiles 3 and 4, and quartiles 1-3 were compared to quartile
4. As shown in Figures 5 and 6, three separate bivariate logistic regression models were
calculated for the assessment of race and gender. In the analysis of race, Caucasians
were used as the referent category. Odds ratios varied from .24 to .91 in Afiican
Americans. The fact that these odds ratios differ across cut points indicates that the
proportional odds assumption was not met (Figure 5). In the analysis of class
membership, the communication class was used as the referent category. Odds ratios
varied minimally from 1.58 to 1.68 which illustrates the proportional odds assumption

was met (Figure 6).

79

Figure 5. Graphical representation of the proportional odds assumption for the
unadjusted analysis of race (African American vs. Caucasian)

Odds Ratio

 

2.00

1.80 a

1.60 -
1.40 i
1.20
1.00

0.80 1 A

0.60 it *

0.40 r- -

0.20

 

0.00

Referent

 

02-4 to Qt 03-4 to 01-2 Q4 to 01-3
Physical Activity Category

 

Figure 6. Graphical representation of the proportional odds assumption for the
unadjusted analysis of class membership (Healthy Lifestyles vs. Communications)

Odds Ratio

2.40

 

2.20 ..

2.00 -

1.80 tr

1.60 H

1.40

1.20

1.00 -

0.801-

 

 

 

 

 

 

0.60

Referent

1

02-4 to Q1 03-4 to Q1-2 Q4 to 01-3
Physical Activity Category

80

4.6 Study Aim 1 Bivariate Ordinal Logistic Regression Results

To quantify the magnitude of the relationships between the exposure variables
listed in section 3.9 and physical activity, we used bivariate ordinal logistic regression to
generate unadjusted odds ratios for quartiles of physical activity at the beginning of a
semester (Table 4). These results illustrate that the odds of being in a higher quartile of
physical activity are 1.83 times greater in males compared to females. This odds ratio
applies to any adjacent quartile (i.e., Q1 compared to Q2-4, Q1-2 compared to Q3-4, or
Q1-3 compared to Q4) and is the universal odds ratio used to describe the magnitude of
the relationship between gender and physical activity across each dichotomization.
Subjects enrolled in the healthy lifestyles class have 62% higher odds of belonging to a
higher quartile of physical activity than those enrolled in the communications class.
Kinesiology majors are 71% more likely belong to a higher quartile of physical activity
than non-kinesiology majors. African Americans have a 64% lower odds of belonging to
a higher quartile of physical activity than Caucasians. When compared to 18 years olds,
subjects 22 years of age or older were 48% less likely to belong to a higher quartile of
physical activity, 21 year-olds were 36% less likely to belong to a higher quartile of
physical activity, and 19 year-olds were 32% likely to belong to a higher quartile of
physical activity. Students living on-campus were 29% more likely to belong to a higher
quartile of physical activity. When compared to students who participated in no high
school sponsored sports, students who participated in 4 or more sports were 2.7 times
more likely to be in a higher physical activity category in college. Those playing 3
sports, 2 sports, or 1 sport during high school were 96%, 61%, and 28% more likely to be

in a higher physical activity quartile during college. When compared to students who

81

were categorized as sedentary with respect to participation in high school LTPA, those
who were categorized as vigorous were 3.4 times more likely to be in a higher physical
activity quartile. Those who were categorized as moderate were 3.8 times more likely
than the sedentary group to be in a higher physical activity category during college. As
stated earlier, Table 4 includes the bivariate unadjusted odds ratios and the results for the
proportional odds assumption when using ordinal logistic regression. Because gender,
race, and BMI do not meet the proportional odds assumption, the final multivariate
ordinal logistic regression model that addressed the relationship between the exposure
variables also did not meet this assumption. As a result, a multinomial logistic regression
was used in place of ordinal logistic regression to assess the relationship between the
exposure variables and quartiles of physical activity for all subsequent analyses for Study

Aim 1 and is described in more detail in section 4.6.

82

Table 4. Unadjusted ordinal logistic regression results for quartiles of physical activity at

the beginning of a semester

 

 

Variable OR 95% Cl's p-value Test for parallel lines
Gender
Male 1.83 1.57 2.12 0.000 0.030
Female 1.00
Class
Healthy Lifestyles 1.62 1.28 2.05 0.000 0.954
Communications 1 .00
Major
Kinesiology 1.71 1.43 2.03 0.000 0.643
Other 1.00
Race
African American 0.36 0.23 0.56 0.000 0.000
Other 0.60 0.39 0.91 0.017
Caucasian l .00
Age
22 or older 0.52 0.34 0.79 0.002 0.838
21 0.64 0.44 0.93 0.020
20 0.75 0.53 1.07 0.112
19 0.68 0.49 0.94 0.020
18 1.00
Living Status
On-campus 1.29 1.02 1.63 0.035 0.267
Off-campus 1 .00
Television viewing
4 or more hours 0.65 0.42 1.02 0.059 0.728
3 hours 0.75 0.53 1.08 0.126
2 hours 0.83 0.63 1.10 0.188
less than 2 hours 1.00
Video game use
0 hours 0.85 0.53 1.37 0.513 0.095
1 hour or less 1.59 0.95 2.66 0.078
2 or more hours 1.00
BMI
Obese 0.86 0.51 1.44 0.564 0.019
Overweight 1.1 1 0.83 1.50 0.473
Normal weight 1.00
High school sports
4 or more 2.71 1.74 4.20 0.000 0.733
3 sports 1.96 1.35 2.86 0.000
2 sports 1.61 1.11 2.33 0.012
1 sport 1.28 0.87 1.88 0.218
0 sports 1.00
High school LTPA
Vigorous 3.41 2.51 4.64 0.000 0.179
Moderate 3.87 2.42 6.20 0.000
Inactive 1.16 0.86 1.57 0.326
Sedentary 1.00

 

83

4.7 Study Aim 1 Bivariate Multinomial Logistic Regression Results

A multinomial regression is an extension of bivariate logistic regression and is
used when the outcome variable has three or more categories. In contrast to the ordinal
logistic regression, multinomial logistic regression does not treat the outcome variable as
ordinal. As a result, specific odds ratios are calculated to describe the magnitude of the
relationship between the exposure variables and each category of the outcome variable.
Table 5 illustrates the results for the unadjusted multinomial logistic regression. In this
analysis, the lowest physical activity quartile was used as the referent group. The three
variables that violated the proportional odds assumption of ordinal regression were
gender, race, and age. The bivariate multinomial regression results for these three
variables indicate that when compared to the lowest physical activity quartile, males had
2.2 higher odds of being in the second quartile than females, 2.2 higher odds of being in
the third quartile, and 5.3 higher odds of being in the fourth quartile. When compared to
Caucasians, African Americans were 80% less likely to be in the second quartile, 85%
less likely to be in the third quartile, and 62% less likely to be in the fourth quartile.
When compared to normal weight students, those who were overweight had the same
odds of belonging to quartile 2, 42% lesser odds of belonging to quartile 3, and 37%
greater odds of belonging to quartile 4.

The bivariate multinomial logistic regression results for the remaining variables
were as follows: Students enrolled in the healthy lifestyles class were 1.56 times more
likely than students enrolled in the communications class to be in the third quartile and 2
times more likely to be in the fourth quartile. Kinesiology majors were 1.82 times more

likely than non-majors to be in the second quartile, 2.3 times more likely to be in the third

84

quartile, and 2.5 times more likely to be in the fourth quartile. When compared to 18
year olds, students who are 22 years or older are 78% less likely, 21 year olds are 69%
less likely, and 19 year olds are 64% less likely to be in the fourth quartile. When
compared to participating in zero high school sponsored sports, those individuals
participating in 4 or more sports were 2 times more likely to be in either the second or
third quartile and 4 times more likely to be in the fourth quartile during college. Students
participating in either two or three sports during high school were also more likely to
belong in physical activity quartile 2, 3, or 4 during college. Compared to the lowest
physical activity quartile, students who were vigorously active in non-sponsored high
school physical activities were 1.83 times more likely to be in the second physical
activity quartile, 4 times more likely to be in the third quartile, and 5.4 times more likely
to be in the fourth quartile of physical activity during college. Students who were
classified as moderately active during non sponsored high school activities were 2.5
times more likely to be in the second quartile, 3.8 times more likely to be in the third
quartile, and 7 times more likely to be in the fourth quartile.

The results Show that healthy lifestyles students were more likely than
communication students to be physically active. In addition, kinesiology majors were
more likely than non-kinesiology majors to be physically active. AS shown in Table 2,
only four kinesiology majors were enrolled in the communications class compared to 189
in the healthy lifestyles class. Therefore, it is possible that differences in physical activity
participation between classes are due to the large percentage of kinesiology majors
enrolled in the healthy lifestyles class. In order to test for this possibility, unadjusted

bivariate logistic regression was used to determine the magnitude of the association

85

between non-kinesiology majors enrolled in the healthy lifestyles class and the
communications class. The results showed that when using the first quartile of physical
activity as the referent group, the odds (95% CIs) belonging to quartile 2, 3, or 4 were
similar (1.0 (66,151); 1.2 (.79, 1.83); 1.2 (80,190), respectively) between these two
groups. In addition, when compared to non-kinesiology majors enrolled in the healthy
lifestyles class, the odds of kinesiology majors belonging to quartiles 2, 3, and 4 were 1.9
(1.06, 3.57); 2.2 (1.22, 3.99); 3.55 (1.99, 6.31), respectively. This shows that the large
number of kinesiology majors enrolled in the healthy lifestyles class was primarily
responsible for differences in physical activity participation between students enrolled in
the two classes at the beginning of the semester. Therefore, the class membership

exposure variable was omitted when conducting the multinomial logistic regression.

86

Table 5. Unadjusted multinomial logistic regression results

 

 

OR OR OR
Q2 to Q3 to Q4 to
Variable Q1 95 % CI Q1 95 % CI Q1 95 % CI
Gender
Male 2.17 1.41 3.33 2.18 1.42 3.36 5.30 3.47 8.08
Female 1 .00 1.00 1 .00
Class
Healthy Lifestyles 1.22 0.84 1.77 1.56 1.08 2.26 2.06 1.42 3.00
Communications 1.00 1 .00 l .00
Major
Kinesiology 1.82 1.07 3.1 1 2.34 1.39 3.93 4.12 2.50 6.79
Other 1.00 1.00 1.00
Race
African American 0.20 0.09 0.42 0.15 0.06 0.33 0.38 0.21 0.70
Other 0.73 0.40 1.33 0.37 0.18 0.75 0.56 0.29 1.06
Caucasian 1 .00 1 .00 1.00
Age
22 or older 1.16 0.61 2.21 0.60 0.30 1.18 0.41 0.21 0.83
21 1.00 0.55 1.82 0.66 0.36 1.21 0.56 0.31 1.02
20 1.04 0.59 1.86 1.00 0.57 1.73 0.65 0.37 1.14
19 1.01 0.59 1.70 0.77 0.46 1.28 0.60 0.36 0.99
18 1.00 1.00 1.00
Living Status
On-campus 0.86 0.59 1.24 1.10 0.76 1.59 1.40 0.97 2.03
Off-campus l .00 1.00 1 .00
Television viewing
4 or more hours 0.63 0.32 1.23 0.49 0.24 0.99 0.56 0.28 1.1 l
3 hours 0.71 0.41 1.24 0.53 0.30 0.96 0.70 0.41 1.22
2 hours 0.91 0.59 1.42 0.84 0.54 1.30 0.74 0.47 1.17
less than 2 hours 1.00 1.00 1.00
Video game use
0 hours 1.26 0.60 2.65 1.66 0.74 3.70 0.72 0.36 1.42
1 hour or less 2.41 1.04 5.58 2.96 1.21 7.23 2.20 1.02 4.75
2 or more hours 1.00 1.00 1.00
BMI
Obese 0.72 0.32 1.60 0.72 0.33 1.57 0.80 0.36 1.80
Overweight 1.00 0.63 1.58 0.58 0.35 0.97 1.36 0.87 2.12
Normal weight 1 .00 1 .00 1.00
High school sports
4 or more 2.06 1.00 4.26 2.14 1.04 4.41 4.57 2.26 9.26
3 sports 1.80 1.00 3.22 2.18 1.23 3.86 2.75 1.50 5.04
2 sports 2.05 1.17 3.59 1.86 1.06 3.28 2.16 1.18 3.94
1 sport 1.44 0.81 2.57 1.16 0.64 2.09 1.55 0.83 2.87
0 sports 1.00 1.00 1.00
High school LTPA
Vigorous 1.83 1.10 3.07 4.15 2.51 6.88 5.39 3.27 8.88
Moderate 2.49 1.04 5.96 3.81 1.60 9.07 7.18 3.18 16.20
Inactive 1.62 1.05 2.48 1.66 1.04 2.64 1.04 0.62 1.73
_ Sedentary 1.00 1.00 1.00

 

87

4.8 Study Aim 1 Adjusted Multinomial Logistic Regression Results

Table 6 shows the final multivariate multinomial logistic regression model that
addresses the relationship between gender, major, high school sponsored sport and LTPA
participation and quartile of physical activity. Gender, race, age, and major were forced
into the model. Television, video game use, BMI category, high school sport
participation, and high school LTPA were entered stepwise into the model. Of these,
only high school sport and high school LTPA remained significant and were retained.
When compared to the first physical activity quartile, males had 2.5 greater odds of
belonging to the second quartile than females, 2.6 greater odds of belonging to the third
quartile, and 7.2 greater odds of belonging to the fourth quartile. After adjustment for the
exposure variables, the odds of males being in higher physical activity quartiles remained
greater than females. The odds of belonging to quartile 4 increased from 5.3 to 7.2 after
adjustment. Figure 5 illustrates the cumulative change in odds ratios (quartile 4 to quartile
1) for gender following adjustment for the exposure variables included in the model.
Students who were kinesiology majors were 2 times more likely than communications
students to be in the second quartile, 2.4 times more likely to be in the third quartile and
4.2 times more likely to be in the fourth quartile. After adjustment for the exposure
variables, the results for the impact of major on physical activity quartile were similar to
the unadjusted analysis. Afiican American students were 79% less likely than
Caucasians to be in the second quartile, 87% less likely to be in the third quartile, and
56% less likely to be in fourth quartile. When compared to 18 year olds, students who
were 19 or 20 years of age were 55% less likely to be in the highest physical activity

quartile. Students who were 22 years or older were 62% less likely to be in the highest

88

quartile. Students who participated in 4 or more high school sports were 2.2 times more
likely to be in the second physical activity quartile, 2 times more likely to be in the third
quartile and 5.2 times more likely to be in the highest quartile during college, compared
to those who participated in zero high school sports. Those who participated in 3 Sports
were 1.7, 2.1, and 3.1 times more likely to be in the second, third, and fourth quartile
during college, respectively. Students who participated in 2 sports during high school
were 2.3, 2.2, and 3.3 times more likely to be in the second, third, and fourth quartile
during college, respectively. When compared to students who were classified as
sedentary in high school LTPA, students who were vigorously active were 1.5 times
more likely to be in the second quartile of physical activity, 3.1 times more likely to be in
the third quartile, and 4.3 times more likely to be in the fourth quartile. Students who
were classified as moderately active in LTPA were 2.5 times more likely to be in the
second quartile of physical activity, 3.7 times more likely to be in the third quartile, and
9.9 times more likely to be in the fourth quartile. There were no significant interactions
in the analysis. The global p—value represents the level of significance of the entire

model.

89

Table 6. Adjusted multinomial logistic regression results for quartiles of physical activity
at the beginning of a semester based on high school physical activity

 

 

OR OR OR
Q2 Q3 Q4 Global
to to to p-
Variable Q1 95 % CI Q1 95 % CI Q1 95 % CI value
Gender <.001
Male 2.54 1.59 4.08 2.56 1.59 4.13 7.17 4.37 11.76
Female 1.00
Major
Kinesiology 2.01 1.14 3.56 2.36 1.34 4.15 4.23 2.40 7.47
Other 1.00 1.00
Race
Afiican
American 0.21 0.09 0.47 0.17 0.07 0.43 0.44 0.21 0.96
Other 0.78 0.40 1.52 0.44 0.20 0.96 0.70 0.32 1.51
Caucasian 1 .00 1 .00 1 .00
Age
22 or older 1.35 0.66 2.76 0.73 0.34 1.57 0.38 0.16 0.88
21 years 1.13 0.59 2.13 0.73 0.38 1.42 0.72 0.37 1.43
20 years 0.84 0.46 1.56 0.86 0.47 1.56 0.46 0.24 0.88
19 years 0.89 0.51 1.54 0.69 0.40 1.20 0.44 0.24 0.79
18 years 1.00 1.00 1.00
High school
sports
4 or more 2.17 0.99 4.73 1.97 0.89 4.36 5.24 2.29 11.98
3 sports 1.73 0.92 3.27 2.07 1.10 3.89 3.07 1.50 6.26
2 sports 2.31 1.25 4.25 2.17 1.16 4.04 3.33 1.64 6.76
1 sport 1.59 0.86 2.96 1.30 0.68 2.49 2.12 1.03 4.35
0 sports 1.00 1.00 1.00
High school
LTPA
Vigorous 1.54 0.90 2.66 3.14 1.84 5.35 4.31 2.46 7.55
Moderate 2.51 1.01 6.23 3 .74 1.52 9.22 9.88 4.02 24.27
Inactive 1.52 0.96 2.41 1.38 0.84 2.26 1.03 0.59 1.82
Sedentary 1 .00 1.00 1 .00

 

90

Figure 7. Change in odds ratios for gender after adjustment for exposure variables
included in the adjusted multinomial logistic regression model

 

 

 

 

 

 

 

 

 

 

12 ..
111- ~- ------- - ~71» -
10,,, * ----'5~er- L- —
a 71 L
m
8 5.
a 5,, e
4 L
3, L
2 ._.
1 r r
Referent Gender ijor Race Age HS sports HS LTPA
Variables Included in Adjusted Model
4.9 Description of Study Aim 2

The second Study Aim assessed the impact of college students’ enrolhnent in a

semester long healthy lifestyles class on change in physical activity participation. AS

explained in section 4.2, an underlying objective of this research was to assess if the

impact of the exposure variables change when using different physical activity outcome

measures. As we found in Study Aim 1, the results were similar when using the different

physical activity outcome variables. Therefore, we also used the ordinal specification of

physical activity (i.e., quartiles) in Study Aim 2 (see section 4.3). The research question

for Study Aim 2 was:

Study Aim 2: Does enrollment in a 3-month long healthy lifestyles class

influence changes in physical activity energy expenditure?

91

 

4.10 Time Varying Covariates

It is of interest to consider whether the impact of enrollment in the healthy
lifestyles class on physical activity participation is different with respect to the time
varying covariates (BMI, television viewing, video game use, living status change). To
assess the change in these variables over the semester, differences from the baseline and
follow-up surveys were calculated. Due to the brief duration of the study period, changes
in BMI (.25i.85 units), television viewing (-.01i.96 hours) and video game use (.01i.59
hours) were negligible. Therefore, it is unlikely that the time varying covariates are
contributing to change in physical activity. To test for this, we used three level
categorical variables to assess the change in BMI, television viewing, and video game use
in all subsequent analysis. Only 19 students reported a change in living status during the
semester. AS a result, the baseline variable for living status was used in all subsequent

analyses. These variables are described in more detail in section 3.3.

4.11 Study Aim 2 Descriptive Results

Chi square analysis was used to assess the relationship with each predictor
variable and the change in physical activity during enrolhnent in the healthy lifestyles
class or communications class. Change in physical activity was defined as a three-level
categorical variable: 1) Decline: participants who moved to a lower physical activity
quartile or remained in the lowest physical activity quartile, 2) Stay the same: study
participants who remained in the second, third, or fourth quartile throughout the semester
period, 3) Increase: study participants who increased into a higher physical activity

quartile. The descriptive results for change in physical activity quartile and their

92

associations with the predictor variables are shown in Table 7. There were no significant
relationships between the time variant covariates and change in physical activity. Among
the time invariant covariates, significant relationships were found between change in
physical activity and class membership and major. These statistically significant
associations illustrate that the percentage of students enrolled in the healthy lifestyles
class who increased physical activity (23%) was greater than those who increased in the
communications class (14%). The percentage of females who increased physical activity
participation during the semester (21%) was higher than that of males (14%). In addition,
physical activity energy expenditure decreased an average of 2.1 kcal/kg/week in the
healthy lifestyles during the semester class compared to 6.7 kcal/kg/week in the
communications class. When separating the two healthy lifestyles classes, physical
activity energy expenditure decreased 4.3 kcal/kg/week among students enrolled during
the fall semester and remained the same among students enrolled in the spring semester.
The difference in change in physical activity energy expenditure between the two healthy

lifestyles classes is addressed in supplemental statistical analyses (Section 4.14).

93

Table 7. Descriptive data for change in physical activity quartile

 

 

stay the
decrease same increase
chi-square p-

Variable N N % N % N % value
Gender
Males 255 97 38% 122 48% 36 14% 0.043
Females 508 160 31% 242 48% 106 21%
Class

Healthy Lifestyles 368 115 31% 168 46% 85 23% 0.008
Communications 395 142 36% 196 50% 57 14%

Major
Kinesiology 152 55 36% 67 44% 30 20% 0.593
Other 608 200 33% 296 49% 1 12 18%

Race
African American 46 9 20% 28 61% 9 20% 0.198
Other 63 18 29% 31 49% 14 12%
Caucasian 654 230 35% 305 47% 1 19 18%

Age
22 or older 79 20 25% 38 48% 21 27% 0.404
21 120 39 33% 60 50% 21 18%
20 159 51 32% 79 50% 29 18%
19 211 70 33% 104 49% 37 18%
18 194 77 40% 83 43% 34 18%

Living Status
On-campus 366 132 36% 167 46% 67 18% 0.3 72
Off-campus 393 123 31% 195 50% 75 19%

Television viewing

change
Decrease 205 61 30% 103 50% 41 20% 0.517
Stay the same 346 121 35% 166 48% 59 17%
Increase 1 91 5 8 30% 91 48% 42 22%

Video game use

change
Decrease 83 20 24% 45 54% 18 22% 0.254
Stay the same 570 184 32% 276 48% 110 19%
Increase 89 36 40% 39 44% 14 16%

BM]
Decrease 193 58 30% 97 50% 38 20% 0.881
Stay the same 149 45 30% 75 50% 29 19%
Increase 367 124 34% 172 47% 71 19%

 

94

4.12 Study Aim 2 Unadjusted Bivariate Results

A multinomial logistic regression was used to calculate the unadjusted odds ratios
between exposure variables and change in physical activity across the semester (Table 8).
When compared to individuals who remained in the same quartile of physical activity,
males had 33% lower odds than females to increase physical activity quartile. Students
enrolled in the healthy lifestyles class had 74% greater odds than students enrolled in the
communications class to increase to a higher physical activity category. African
American students had 57% lower odds than Caucasian students to decrease physical

activity quartile during the semester.

95

Table 8. Unadjusted multinomial logistic regression results for change in quartile of
physical activity during a semester.

 

 

decrease vs. increase vs.
stay the stay the

Variable same 95 % CI same 95 % CI

Gender
Male 1.20 0.86 1.68 0.67 0.44 1.04
Female 1.00 1.00

Class
Healthy Lifestyles 0.94 0.69 1.30 1.74 1.17 2.58
Communications 1 .00 1 .00

Major
Kinesiology 1.21 0.82 1.81 1.18 0.73 1.92
Other 1.00 1.00

Race
Afiican American 0.43 0.20 0.92 0.82 0.38 1.80
Other 0.77 0.42 1.41 1.16 0.59 2.25
Caucasian 1 .00 1 .00

Age
22 or older 0.57 0.30 1.06 1.35 0.69 2.63
21 0.70 0.42 1.17 0.85 0.45 1.62
20 0.70 0.44 1.11 0.90 0.50 1.61
19 0.73 0.47 1.12 0.87 0.50 1.50
18 1.00 1.00

Living Status
On-campus 1.25 0.91 1.73 1.04 0.71 1.54
Off-campus 1 .00 l .00

Television viewing change
Increase 0.87 0.58 1.31 1.30 0.81 2.08
Decrease 0.81 0.55 1.20 1.12 0.70 1.79
Stay the same 1.00 1.00

Video game use change
Increase 1.38 0.85 2.26 0.90 0.47 1.72
Decrease 0.67 0.38 1.17 1.00 0.56 1.81
Stay the same 1.00 1.00

BMI
Increase 1.20 0.78 1.86 1.07 0.64 1.78
Decrease 1.00 0.61 1.63 1.01 0.57 1.79
Staythe same 1.00 1.00

 

4.13 Study Aim 2 Adjusted Multinomial Regression Results
Table 9 shows the adjusted odds ratios for change in quartile of physical activity

during a semester enrollment in either a healthy lifestyles class or communications class.

96

Using multinomial logistic regression, gender, class, race, and age were forced into the
model. The remaining time invariant covariates (living status and major) and time
variant covariates (change in BMI, television viewing, and video game use) were entered
stepwise into the model. Due to a lack of statistical significance these variables were not
retained in the model. The results Showed that students enrolled in the healthy lifestyles
class had 1.8] higher odds of increasing physical activity than the students enrolled in the
communications class regardless of gender, age, race, and major. Our results also
showed that the time varying covariates were not associated with changes in physical
activity during the semester. After adjustment for class, age, and race males were 33%
less likely than females to increase physical activity and 22 years olds were 49% less
likely than 18 year olds to decrease their physical activity participation. There were no
significant interactions within the exposure variables. It also should be noted that when
compared to students who decreased physical activity, students enrolled in the healthy
lifestyles class were 2.2 times more likely than communications students to increase
physical activity. After stratifying our sample by class and major, our results also
showed no differences in change in physical activity participation between kinesiology
majors and non-majors enrolled in the healthy lifestyles class. In addition, non-majors
enrolled in the healthy lifestyles class had 1.87 greater odds of increasing physical
activity than non-maj ors enrolled in the communications class. These results imply that
greater improvements in physical activity participation are seen in students enrolled in a

healthy lifestyles class when compared to those who are not.

97

Table 9. Adjusted multinomial logistic regression results for change in quartile of
physical activity during a semester

 

 

decrease increase Global
vs. stay the vs. stay the p-
Variable same 95 % CI same 95 % CI value
Gender 0.112
Male 1.18 0.83 1.70 0.67 0.42 1.06
Female 1 .00 1 .00
Class
Healthy Lifestyles 0.89 0.63 1.27 1.81 1.20 2.73
Communications 1 .00 1 .00
Race
African American 0.58 0.26 1.30 0.74 0.32 1.74
Other 0.65 0.32 1.32 1.11 0.54 2.28
Caucasian 1 .00 1.00
Age
22 or older 0.51 0.25 1.02 1.28 0.63 2.61
21 0.80 0.47 1.37 0.80 0.42 1.54
20 0.71 0.43 1.17 0.89 0.49 1.63
19 0.75 0.47 1.18 0.92 0.52 1.62
18 1.00 1.00

 

4.14 Results for Supplementary Statistical Analyses

A series of supplementary analyses were performed to address the limitations of
the statistical analyses previously described. The first limitation to the prior modeling
strategies was the attrition in subject participation from the baseline survey to follow-up
survey due to the use of a change score. As an alternative, a multivariate model which
included all available physical activity energy expenditure data was used. Table 10
shows the results for the population based model assessing the impact of gender, time,
and class on the log of physical activity energy expenditure throughout the 15-week study
period. The results illustrate that males were more active than females, physical activity
decreased over time, and that physical activity was greater in students enrolled in the

healthy lifestyles class. These results show that the effect of class on physical activity

98

over time is not significant. This suggests the magnitude of the difference in physical
activity at follow-up is not different than at baseline between students enrolled in the
healthy lifestyles and communications classes. This contradicts our original results
which suggest there was a difference between the two classes. In contrast to the
population-based approach, a subject-specific modeling strategy assessing the impact of
gender, time, and class on physical activity throughout the 15-week study period was also
conducted (Table 11). The results did not differ between the population based and
subject-specific approaches.

Table 10. Estimated relationship between physical activity energy expenditure and class
membership during enrollment in a healthy lifestyles class (population-based approach)

 

 

Variable Coefficient 95% CI J

Gender 0.456 0.345 0.565 <.001
Time -0.186 0.268 -0.103 <.001
Class 0.284 0.161 0.407 <.001
Class over Time 0.087 -0.028 3.130 0.139

 

Table 11. Estimated relationship between physical activity energy expenditure and class
membership during enrollment in a healthy lifestyles class (subject-specific approach)

 

 

Variable Coefficient 95% CI p

Gender 0.456 0.345 0.568 <.001
Time -0.183 0.263 -0.103 <.001
Class 0.30] 0.191 0.426 <.001
Class over Time 0.095 -0.020 2.77 0.106

 

The second limitation addressed was the value of the calculated change score. An
alternative to this approach is to use the follow-up energy expenditure value as the
outcome variable and assess the impact of the exposure variables such as gender and
class while holding the baseline energy expenditure value constant as shown in Table 12.

The results show that after controlling for physical activity energy expenditure at

99

baseline, physical activity energy expenditure as follow-up was higher in students
enrolled in the healthy lifestyles class. Similar to the original change score results, this
analysis illustrates that students enrolled in the healthy lifestyles class are more likely to
improve physical activity during the semester.

Table 12. Estimated relationship between physical activity energy expenditure at follow-

up and class membership after controlling for physical activity energy expenditure at
baseline

 

 

Change Score Coefficient 95% CI p-value
Log kcal/kg/week at baseline 0.561 0.504 0.618 <.001
Gender (male) 0.089 -0.022 0.201 0.117
Class (healthy lifestyles) 0.141 0.081 0.201 <.001

 

The third limitation addressed was the potential differences that existed between
different sections of the healthy lifestyles class. Specifically, students enrolled in the
same section within the healthy lifestyles class are more alike when compared to students
enrolled in a different section of the healthy lifestyles class. In order to address this
limitation, a modeling approach that recognized the nested structure of the students
within the healthy lifestyles sections was conducted and is shown in Table 13. These
results illustrate that after controlling for physical activity participation at baseline and
the nested structure of the healthy lifestyles class, physical activity participation at the
end of the semester was higher in the students enrolled in the healthy lifestyles classes.
These results are similar to the results shown in Table 11 which did not control for the

nested structure of the healthy lifestyles class.

100

Table 13. Estimated relationship between physical activity energy expenditure at follow-
up and class membership after controlling for physical activity energy expenditure at
baseline and the nested structure of the healthy lifestyles class

 

 

Change Score Coefficient 95% CI p-value
Log kcal/kg/week at baseline 0.594 0.457 0.731 <.001
Gender (male) 0.105 -0.061 0.271 0.214
Class (healthy lifestyles) 0.145 0.006 0.283 0.041

 

A fourth limitation within the previous analyses was the use of the class
membership variable that did not distinguish between students enrolled in the fall and
spring healthy lifestyles classes. When analyzing the data using a 3-level categorical
variable for class (fall communications, fall healthy lifestyles, spring healthy lifestyles),
results indicate that: 1) males are more active than females, 2) students enrolled in either
the fall or spring semester healthy lifestyles classes are more active than students enrolled
in the communications class, 3) physical activity declines over time, 4) kinesiology
majors are more active then non-majors, 5) Africa Americans are less active than
Caucasians, and 6) those who watch more television are less active (Table 14). These
results also show no difference in physical activity throughout the semester between the
fall healthy lifestyles class and fall communications class. In contrast, when compared to
the fall communications class, physical activity participation was more likely to increase
in students enrolled in the spring healthy lifestyles class. Furthermore, physical activity

participation in males decreases during the semester when compared to females.

101

Table 14. Estimated relationship between physical activity energy expenditure and
enrollment in the fall or springhealthy lifestyles classes

 

 

Variable Coefficient 95% CI p
Gender

Female 1 .000 Reference

Male 0.510 0.376 0.644 <.001
Class

Communications 1 .000 Reference

Healthy Lifestyles Fall 0.244 0.075 0.025 0.005

Healthy Lifestyles Spring 0.233 0.074 0.392 0.004
Time

Baseline l .000 Reference

Follow-up -0.087 -0.194 0.019 0.108
Major

Non-kinesiology major 1.000 Reference

Kinesiology major 0.270 0.112 0.428 0.001
Age -0.045 -0.096 0.006 0.090
Race

Caucasian 1 .000 Reference

African American -0.419 -0.714 -0.123 0.005

Other -0.133 -0.371 0.105 0.273
Television viewing -0.072 -0.121 0.021 0.005
Video game use 0028 -0.103 0.047 0.468
Body Mass Index 0.009 -0.008 0.026 0.281
Living Status

On-campus 1 .000 Reference

Off-campus 0.066 -0.072 0.203 0.349
Class Fall over Time

Communication 1 .000 Reference

Healthy Lifestyles Fall -0.034 -0.176 0.108 0.638
Class Spring over Time

Communication 1 .000 Reference

Healthy Lifestyles Spring 0.159 0.019 0.296 0.026
Gender over Time

Female 1 .000 Reference

Male -0.126 -0.250 -0.003 0.045

 

102

CHAPTER 5
DISCUSSION
5.] Effect of Gender and Class on Physical Activity Participation

The results of this study illustrate that both gender and academic major influenced
physical activity participation in college students. We also found physical activity
participation was lower in older college students and Afiican Americans and as
hypothesized, higher in males. It was also hypothesized that physically active college
students would be more likely to enroll in the healthy lifestyles class, resulting in
selection bias. However, the higher physical activity participation of students enrolled in
the healthy lifestyles class was due to the large number of kinesiology majors enrolled in
that class rather than simply that physically active college students were more likely to
enroll.

The impact of gender on physical activity participation in college students has
been studied previously, but has provided mixed results. Similar to our results , previous
studies have shown that a larger proportion of male college students (45-49%) were more
likely to report regular exercise than female students (35-39%) 48’ 65. Dinger and
Waigandt '9 found that males participated on an average 2.8 days of vigorous physical
activity per week compared to 2.2 days for females. However, other studies have found
no difference in physical activity participation between males and females 44’ 58. A
potential reason for the inconsistencies in these results is that physical activity was
reported differently within each of these previous studies. That is, physical activity has

been reported as the number of days per week of vigorous physical activity, the number

of days of structured exercise, or as a categorical variable based on national physical

103

activity recommendations. Similar to our study, Leslie and associates 43 used a more
detailed ordinal assessment of physical activity, which used frequency, intensity, and
duration of physical activity to determine weekly physical activity energy expenditure
(kcal/week). These authors also reported that females were more likely than males to be
sedentary (<100 kcal/week) or low active (100-799 kcal/week), whereas males were more
likely to be moderately active (>800 kcal/week of non-vigorous physical activity) or
vigorously activity (>1600 kcal/week of physical activity that included vigorous physical
activity).

Selection bias is a potential problem when assessing physical activity behaviors in
college students enrolled in an activity based class as it is possible a more health-
conscious, physically active student is more likely to enroll in such a class. As a result,
their physical activity levels may be different from the general population of college
students. However, no previous study has addressed directly whether this selection bias
exists. Our initial results show that this selection bias exists as students enrolled in the
healthy lifestyles class were more active than students who were not enrolled in this
class. However, the large percentage of kinesiology majors, who are more active than
non-kinesiology majors, were enrolled in the healthy lifestyles class and accounted for
the higher physical activity levels among students enrolled in the healthy lifestyles class.
To illustrate this, when stratifying the sample by class, kinesiology majors enrolled in the
healthy lifestyles class were 1.8 times more likely than non-kinesiology majors enrolled
in the healthy lifestyles class to be in the second quartile, 2.3 times more likely to be in
the third quartile, and 4.1 times more likely to be in the highest quartile. Furthermore,

when comparing the non-kinesiology majors enrolled in the healthy lifestyles class to

104

those enrolled in the communications class, there were no differences in physical activity
participation. These results suggest that when assessing physical activity behaviors in
college students enrolled in an activity based class, it is important to address the potential
selection bias that may exist due to a large percentage of kinesiology majors enrolled in
that class. Our results also suggest that the physical activity selection bias does not exist
among non-kinesiology majors as more physically active non-kinesiology majors were
not more likely to enroll in an activity class. Therefore, after accounting for students who
are kinesiology majors, our results suggest that physical activity participation among
students enrolled in a healthy lifestyles class are generalizable to the entire college
student population.

Physical activity may decline throughout enrollment in college due to the
increasing demands of course work, pressures of gaining independence, stress related to
entering the workforce, or other responsibilities that inhibit time spent participating in
physical activity. Although, the age related decline in physical activity from adolescence
into young adulthood is well documented 9’ '5’ 29’ 46’ 68 there are few studies which have
assessed this decline during enrollment in college. Our results show that older students
were less likely to be physically active. Leslie and colleagues 4' assessed age-related
differences in physical activity participation in nearly 2500 college students. The authors
reported that physical activity declined in males from 75% to 70% to 58% in 18-19 years
olds, 20-24 year olds, and 25-29 year olds, respectively. Females declined from 66% to
56% to 46% in the same age groups. Our results illustrate a similar decline as 32% of 18
years olds were in the highest physical activity quartile compared to 24% of 19 year olds,

23% of 20 years olds, 23% of 21 year olds, and 18% of 22 year olds. Both our data and

105

those of Leslie et al 4'

are limited to cross-sectional comparisons and as a result we were
unable to assess the longitudinal declines in physical activity during college. Future
studies Should focus on assessing physical activity behaviors during undergraduate
enrollment to determine if and when the added demands and pressures that occur during
college contribute to the decline in physical activity.

In the United States, there are substantial racial disparities in health across a
lifespan and physical inactivity may be a contributing factor as African Americans are
less active than Caucasians 30. There is limited research regarding racial/ethnic
differences in physical activity within a college population. In order to contribute to the
limited body of literature in African American college students, Ford and Goode 23
assessed daily physical activity levels in 224 students enrolled in an urban African
American university. The authors reported that only 45% participated in daily physical
activity of which 60% were males. The authors were unable to assess racial disparities
because the sample was entirely Afiican American. Harris and colleagues 29 assessed
longitudinal differences in physical activity participation between Afiican American and
Caucasians from adolescence into young adulthood. In contrast to our results which
show that African Americans are less active, there appeared to be little difference in
physical inactivity between African Americans and Caucasians. A potential reason for
the contrasting results is that our study used a more detailed assessment of physical
activity. In addition, our study focused specifically on college students while Harris et a1

29 did not limit their study sample to college students. Due to the limited number of

study participants who considered themselves Hispanic (n=16), Asian (n=3 7), Pacific

106

Islander (n=2), or American Indian (n=l) we were unable to compare physical activity
participation between these groups.

In summary, our study found that males are more likely than females and
Caucasians are more likely than African Americans to be physically active. In addition,
there appears to be a decline in physical activity in college students with age. Initial
results illustrated that students enrolled in a healthy lifestyles class were more active than
students enrolled in a general communications class. However, the majority of this
relationship was due to large number of kinesiology majors enrolled in the activity class.
That is, non-kinesiology majors enrolled in the activity class had similar levels of
physical activity compared to those not enrolled in the class. Furthermore, non-
kinesiology majors enrolled in the activity class had lower levels of physical activity
compared to kinesiology majors enrolled in the class. Although, physical inactivity is a
major concern on college campuses, there is a lack of literature that describes who is
more likely to be inactive in this population. Our results add to this limited body of
knowledge and enhance our understanding of the determinants of physical activity

participation in the college student population.

5.2 Effect of High School Sports and Other Leisure Time Physical Activities on
Physical Activity in College

The results of our study Show a positive dose response relationship between
number of high school sponsored sports played and physical activity participation in
college. In addition, regardless of high school sponsored sport participation, students

who were either moderately or vigorously active in leisure time physical activities other

107

than school sponsored clubs or sports during high school were more likely to be
physically active in college.

The transition from high school into college is an event which may contribute to
the decline in physical activity. Bray and Born ” compared vigorous physical activity
levels between high school and the first year of college and showed that frequency of
vigorous physical activity declined from an average of 3.3 sessions per week during high
school to 2.7 sessions per week during college. Leighton and Swerissen 40 reported that
42% of college students thought they were less active compared to their last year of high
school. The goal of this study was not to determine the level of decline in physical
activity during the transition from high school to college, but to add to the current body
of literature by assessing the influence of physical activity and sport participation during
high school on physical activity behaviors during college.

Our results show that participation inhigh school sponsored sports influences
physical activity participation during college. Compared to the lowest physical activity
quartile, college students who participated in a greater nrunber of high school sports were
more likely to be physically active in college. Everhart and colleagues 2‘ examined the
impact of participation in high school athletics on physical activity levels in 201 college
students. They concluded that physical activity participation during college was not
different between former high school athletes and non-athletes. In comparison to the
previous study which dichotomized high school sport participation (yes/no), an advantage
to our study is athletes were categorized based by the number of high school sports

played. This offered a more detailed assessment of high school sport participation

108

allowing us to uncover the positive dose response relationship between high school sport
participation and college physical activity participation.

It is also important to assess the impact of high school physical activity
participation outside of school sponsored sports or club sports as it may also have an
impact on physical activity during college. Previous research has shown that physical
activity participation tracks moderately (r=.57) from high school into college '0.
However, previous studies have not distinguished between high school sport and leisure
time physical activity. Our results also show that regardless of participation in high
school sports, there is a positive dose response relationship between high school LTPA
and physical activity during college. These results imply that any physical activity
participation, even outside of high school sponsored sports, has a beneficial effect on
physical activity levels during college. This finding supports the value of encouraging
the promotion of leisure time physical activity to high school students via avenues such
as school wellness programs and physical education classes.

While the results of our study offer new insights on the relationship between
physical activity participation in high school and college, there are limitations to this
study. It is possible that sport and physical activity participation during a student’s senior
year of high school may have a more direct impact on physical activity participation
during college. However, it was not possible to assess physical activity participation
during a student’s last year of high school as our survey focused on estimating high
school physical activity and sport participation across all four years of high school. We
also did not assess attitudes towards and barriers to physical activity as previous research

has shown they are potential determinants of physical activity during the transition from

109

high school to college '0’ H

. F urtherrnore, when assessing high school leisure time
physical activities, we did not specifically assess participation in physical education
classes, after school wellness programs, or other specific modes of physical activity
promotion. Therefore, we were unable to account for the direct influence of these factors
on college physical activity participation.

In conclusion, our results suggest that there is a dose response relationship
between participation in high school Sports and physical activity participation in college.
Furthermore, our results suggest that regardless of high school sport participation, college
students who are more active in non-sponsored high school activities are more active

during college. This finding illustrates the importance of promoting both high school

sport and physical activity participation to high school students.

5.3 Efl'ect of Enrollment in a Healthy Lifestyles Class on Change in Physical Activity
The purpose of this aim was to assess the impact of enrolhnent in a healthy
lifestyles class, designed to promote and educate college students about the importance of

physical activity and other health related behaviors on physical activity participation.
Each healthy lifestyles class was held twice per week and included a 50 minute lecture on
a health related topic followed by a 50 minute planned activity designed to promote the
participation in physical activity. A communications class was used as a comparison
group as the students enrolled were not receiving the same education about physical
activity. The results of our study showed that during a 15-week semester, students
enrolled in the healthy lifestyles class were more likely to increase physical activity

compared to students enrolled in the communications class.

110

The majority of research assessing physical activity participation in college
students enrolled in activity based classes are cross-sectional and Show that 23-58% of
those students are physically active 6’ 12’ 23‘ 5 1’ 56. Our baseline results showed that 63% of
our population was classified as vigorously activity and 15% as moderately active which
is slightly higher than previous reports. However, due to the different methodologies
used to assess physical activity, it is difficult to make a direct comparison of our results
with the previous literature. Furthermore, the change in physical activity during
enrolhnent in the class could not be assessed as the previous studies did not include a
follow-up assessment of physical activity.

Our study, which included the assessment of physical activity at the beginning
and end of enrolhnent in the two classes provides insight on the impact of enrollment in a
healthy lifestyles class designed to promote physical activity participation compared to
the communications class. Our results showed that 23% of students enrolled in the
healthy lifestyles class increased physical activity during the semester compared to only
14% of those enrolled in the communications class. In addition, 36% of communications
students decreased physical activity during the semester compared to 31% of healthy
lifestyles students. Furthermore, changes in physical activity as a result of the healthy
lifestyles class were similar for kinesiology majors and non-kinesiology majors.

The effectiveness of classroom-based physical activity interventions in college
students is uncertain. To the best of our knowledge, results fiom only two previous
intervention studies have been published in college students 42’ 6’. Project GRAD was
developed to promote the adoption and maintenance of physical activity among young

adults transitioning from the university into adult roles. Three hundred and thirty-eight

111

college seniors from a large urban university in the United States were randomized into
either a class designed to promote physical activity or a control class which covered
general health topics. Each subject completed a 7-day physical activity recall at the
beginning and end of the 15-week classroom based intervention. Results showed the
intervention had no Significant effects on total leisure time physical activity participation
in men. However, in women, the intervention improved total leisure time physical
activity and increased participation in flexibility and strengthening exercises 6'. Leslie
and colleagues 42 assessed the impact of a classroom based intervention at two urban
university campuses. A physical activity program was implemented at one campus over
an 8 week period and included offering students fitness assessments and a free activity
class of their choice. Physical activity was assessed via a self-reported questionnaire
given at the beginning and end of the 8-week period. The authors reported that students
who received the intervention were more vigorously active 42. The results of our study
contribute to the limited research on classroom based interventions in college students as
we found students enrolled in the activity based class were more likely to increase
physical activity participation than those enrolled in a non-activity based class. In
contrast to Project GRAD 6' which included only college seniors (i.e., older students), our
study included mostly freshmen and sophomores (roughly 50% of our sample were 19
years or younger). Furthermore, Leslie et al 42 limited their intervention period to 8
weeks, whereas, our study spanned the entire duration of a 15-week semester.
Limitations to our study include the duration of the study was only 15 weeks. In
addition, only two time-points were used to assess the change in physical activity over

the semester. This study was observational as we were unable to randomize students into

112

the healthy lifestyles and communications classes. In addition, we did not assess
variables such as attitudes towards physical activity and readiness to begin a physical
activity program which could partially explain why some students enrolled in the healthy
lifestyles class were more likely to increase physical activity participation.

In conclusion, results of our study show that both male and female students
enrolled in an activity based class are more likely to increase physical activity
participation during a 15-week semester. Although these results illustrate that an activity
based class may serve as an opportunity to promote and increase physical activity
participation in college students, there is minimal research confirming these findings.
Therefore, our study provides valuable insight into the short term impact of enrollment in

a healthy lifestyles class.

5.4 Assessment of the Effect of Using Different Specifications of Physical Activity
Energy Expenditure

Physical activity assessment is dependent on frequency, intensity, and duration of
the activity. However, the current literature that assesses physical activity participation in
college students lacks consistency in the use of these factors 36. In order to help address
this issue, an underlying aim of this study was to determine if results changed when
different specifications of physical activity were used. Therefore, physical activity
participation was assessed as a continuous variable (kcal/kg/wk), an ordinal variable
(quartiles of kcal/kg/wk), and a categorical variable (meeting the national
recommendations for physical activity). Final multivariate results were similar regardless

of which outcome was used. Each outcome indicated that males, kinesiology majors, 18

113

year olds, Caucasians, and students who were active in high school were more likely to
be physically active in college.

The majority of previous studies examining physical activity in college students
used the criteria of meeting the national recommendations for physical activity I" 26’ 44' 48‘
57. For example, meeting the recommendations for vigorous physical activity is
considered engaging in physical activity that makes an individual sweat or breath hard on
three or more days per week for 20 minutes per occasion. Although this variable is
derived using a combination of frequency, intensity, and duration of physical activity, the
problem within this approach is the inability to differentiate between those who
participate in substantial amounts of vigorous physical activity and those who barely
meet these recommendations. In order to assess the entire spectrum of physical activity,
a continuous physical activity variable (kcal/kg/wk) can be calculated using a
combination of frequency, intensity, and duration of physical activity. A limitation
associated with measuring physical activity as a continuous variable is that the variable is
unlikely to be normally distributed. As shown in the results of this study (Figure 4), our
data are positively skewed due to the disproportionate number of study participants at the
lower end of the distribution. To successfully normalize the data, a log transformation
was needed. However, transforming the data increased the difficulty of interpreting the
results as physical activity participation was reported in terms of log kcal/kg/wk. After
carefirl analysis, we concluded the ordinal physical activity variable was the most
favorable variable to use for our study. Quartiles of physical activity were derived from

the continuous outcome variable which allowed for the assessment of levels of physical

activity participation without the need for transforming the data.

114

The ordinal outcome variable allowed us to use ordinal logistic regression for data
analysis. Ordinal logistic regression is a preferred method of regression analysis with
ranked data because it calculates a single odds ratio that describes the magnitude of the
relationship across different dichotomizations of the data (e.g. quartile 1 compared to
quartiles 2-4 combined, quartile 1-2 compared to quartile 3—4, and quartile 1-3 compared
to quartile 4) 47. This allows for less cumbersome interpretation. However, the
underlying proportional odds assumption of ordinal logistic regression assumes the odds
ratio is consistent across each dichotomization of the ordinal data. If this assumption is
not met, the single odds ratio that is used to describe the data can not be used to
adequately describe the magnitude of the association. As a result, a different regression
technique such as multinomial logistic regression is needed.

In our study, gender, race, and BMI did not meet the proportional odds
assumption and as a result the odds ratios derived from the ordinal logistic regression
models were not appropriate. Therefore, we used a multinomial regression in place of the
ordinal logistic regression. A multinonrial logistic regression is used for categorical data
that is not ranked. As a result, it does not require the proportional odds assumption. A
multinorrrial regression determines the odds ratios for each category compared to a
designated referent category. In our study the referent category was the lowest quartile of
physical activity. The limitation to this method is that it produces more odds ratios than
the ordinal regression. However, upon closer examination of our data, we found that the
multinomial logistic regression provided a more detailed assessment of the data that the
ordinal regression would have missed. For example, the bivariate unadjusted odds ratios

generated by ordinal logistic regression indicated that college students participating in

115

four or more high school sports were 2.7 times more likely to be in a higher physical
activity quartile. When analyzing the same data using multinomial logistic regression a
dose response relationship was uncovered that was missed by ordinal regression. That is,
when compared to the lowest physical activity quartile during college, students who
participated in 4 or more high school sports were 2 times more likely than those who
played zero high school sports to be in the second quartile, 2.1 times more likely to be in
the third quartile and 4.5 times more likely to be in the highest quartile. Thus, the
multinomial model showed a range of odds ratios between 2.0 and 4.5, whereas the
proportional odds were 2.7.

In conclusion, the use of a categorical, ordinal, or continuous outcome variables
provide similar results when describing physical activity participation in college students.
However, researchers should consider the potential limitations when using these
variables. Using a categorical outcome variable that assesses meeting the
recommendations for physical activity fails to distinguish between college students who
participate in high amounts of vigorous physical activity and those who do not. When
using the continuous variable, it is a possibility that a large number of college students
will be on the lower end of physical activity participation causing a non-normal
distribution. Although our results showed that a log transformation successfully
normalized that data, this complicates the interpretation of the outcome variable.
Therefore, we suggest the use of an ordinal variable to describe physical activity
participation in college students due to its ease of interpretability and ability to assess

levels of physical activity participation.

116

5.5 Assessment of Supplemental Analysis Results on the Impact of Enrollment in a
Healthy Lifestyles Class

A series of supplemental analyses were conducted to address the limitations of the
modeling strategies used in the primary analysis of this project. The goal of the
supplemental analyses was to offer additional insights into our primary results and
attempt to provide alternate interpretations of the impact of enrollment in the healthy
lifestyles class on physical activity.

One limitation of the primary modeling strategies was loss-to-follow-up during
the 15-week study period as these strategies assess data only from those who completed
both the baseline and follow-up surveys. To address this limitation, a modeling strategy
was used that included values for all study participants regardless of participation in the
follow-up survey. Results of this supplementary analysis differed from our original
analysis as it revealed that enrollment in the healthy lifestyles class did not influence
physical activity participation over the 15-week study period. This supplemental analysis
was conducted on the level of the population and assessed the impact of the healthy
lifestyles class on students enrolled in the class as a whole. A possible way to offer a
more detailed assessment of the impact of the healthy lifestyles class was to assess if any
differences existed when using a subject-specific based modeling strategy. That is, to
determine if the effect of the healthy lifestyles class different when considering the level
of the individual as compared to the population. Although there is a trend that enrolhnent
in the healthy lifestyles class improves physical activity at the individual level, the p-
value (p = .102) was not significant. These results suggest the positive impact of the

healthy lifestyles class on physical activity shown in our original analyses may in part be

117

due to baseline differences between the two classes that cannot be entirely controlled by
adjusting for a variety of covariates.

A second modeling strategy used as an alternative to the change score approach as
its calculated value may be misleading. The results of this analysis, which used physical
activity at the end of the semester as the outcome variable regressed upon the exposure
variables which included physical activity at baseline, showed that healthy lifestyles
students were more physically active at the end of the semester. However, similar to our
original analyses, the limitation of this modeling strategy was that it did not account for
the loss-to-follow-up as only 765 subjects were analyzed.

Each semester there are six sections of healthy lifestyles classes which are taught
by different instructors. Each section includes approximately 50 of the nearly 300
students enrolled in the healthy lifestyles class per semester. Because of the nested
structure of this class, it was necessary to assess if any differences in physical activity
existed between the different sections. As expected, physical activity was not altered by
the section of a healthy lifestyles class.

It was also important to assess whether physical activity during the 15-week study
period differed between students enrolled in either the fall or spring healthy lifestyles
class. Our results showed that when compared to the students enrolled in the
communications class, students enrolled in the healthy lifestyles class during the spring
semester were more likely to improve physical activity whereas, students enrolled in the
healthy lifestyles class during the fall were not. These results suggest one of two
possibilities. First, there are unique characteristics among students enrolled in the spring

healthy lifestyles class which makes them more likely to improve physical activity as a

118

result of the class. The only difference we found at baseline between the fall and spring
healthy lifestyles classes was a higher percentage of non-kinesiology majors were
enrolled during the spring semester. This difference may imply that more non-
kinesiology majors enroll into the spring healthy lifestyles class with the goal of making
improvements in physical activity. Second, there is a seasonal effect on physical activity
as students are more likely to become more active during the spring semester when
compared to the fall semester. A limitation is that we do not have data from a spring
communications class in which we could specifically assess differences in physical
activity between and within the different semesters. In addition, we do not have data
assessing barriers to physical activity that may change during a semester or the readiness
students to change physical activity during the semester.

Our analysis also showed that males were more likely to decrease physical
activity during the semester than females. Previous research in college students has
Shown that females are more likely to improve physical activity as a results of a physical
activity interventions 6'. Therefore, it is possible that females are more likely than males
to improve physical activity as a result of enrolhnent in the healthy lifestyles class.
However, an additional supplemental analysis showed that impact of enrollment in the
healthy lifestyles was Similar within gender. Therefore, it is likely that males decrease
physical activity because they are more active than females at the beginning of the
semester.

In conclusion, the supplemental analyses preformed provide additional insight
into the initial results of this study. These results contradicted our original findings in

that physical activity does not appear to be influenced by enrolhnent in a healthy

119

lifestyles class. In addition, a potential seasonal effect on physical activity was revealed
as students enrolled in the spring healthy lifestyles class were more likely to improve

physical activity during a lS-week semester.

5.6 Future Studies of Physical Activity in College Students

The third aim of this study was to develop a rationale/methodology for a future
study assessing physical activity participation in college students. As stated throughout
the literature review and discussion of this study, there is a lack of research assessing
physical activity participation in college students. The majority of physical activity
research in college students is cross-sectional 6’ ”’ '2’ 28’ 43’ 58. A recent meta-analysis
indicated that research on college student physical activity is limited by: 1) the lack of
studies in this population, and 2) inconsistent methodology used to assess physical
activity 36. After the comprehensive literature review performed in this dissertation and
the information gained from the results of this study, we concluded there is a need to
assess physical activity and health behaviors of college students across and beyond the
four years of enrollment in a university.

Our findings provide interesting insight on the future direction of physical activity
studies in college students. We were able to uncover important determinants of physical
activity in college students such as gender, age, race, and major. Although this
information contributes greatly to the limited body of knowledge regarding physical
activity in college students, it does not address questions such as “Why are males more
active than females?”, or “Why are older college students less active?”. As a result,

future research should attempt to answer these questions by assessing attitudes towards

120

physical activity, readiness to participate in physical activity, self efficacy of physical
activity, and barriers to physical activity.

It is also important for future studies to assess physical activity and health
behaviors throughout and beyond enrollment at a university. First, our results illustrate
that enrollment in a healthy lifestyles class may benefit physical activity participation
over a 15-week period. To determine if there is long-terrn impact of enrolhnent in the
healthy lifestyles class, there is a need to follow these students beyond 15-weeks and
throughout enrollment in the university. Second, our results suggest an age related
decline in physical activity during enrolhnent in college. By assessing college students
longitudinally throughout four years of university enrolhnent, it is possible to uncover the
determinants for this age related decline. In addition, it would be possible to determine if
and when there is a critical time period in which physical and health behaviors decline.
Third, future studies should also focus on following college students beyond enrollment
at a university and into adulthood. By following these students into adulthood,
researchers could determine if changes in health behaviors during college affect lifelong
physical activity participation.

Future studies Should include a detailed assessment of physical activity similar to
what was used in this study. Specific questions should be asked about intramural sport
participation during college, club and/or varsity Sport participation, and enrollment in
activity/sport based classes to determine their impact on physical activity participation
during college. A detailed assessment of physical activity in combination with questions
aimed at determining why there are differences in physical activity among college

students will add to the existing physical activity literature in this population. The

121

methodology of a study designed to expand the results of this study and further address
the limitations of the current body of literature addressing physical activity participation

in college students is shown in Appendix F.

5.7 Conclusion

AS stated by Healthy Campus 2010, physical inactivity is a major concern among
college students 5. Despite this concern, there is a lack of physical activity data in this
population 36. Therefore, the results of this study add significantly to the body of
literature as we have shown various determinants of physical activity in college students
and the positive impact of a healthy lifestyles class on physical activity participation. In
addition, our study has also provided the foundation for the methodology needed in future

studies.

122

Appendix A

Consent Form
Dear Student:

Researchers from Michigan State University would like you to answer some questions about your physical
activity and health behaviors through an intemet survey. To protect your privacy, all information you
provide will remain confidential. You will have one-week from the in class presentation to complete the
baseline survey. Please take you time and fill out the survey completely. It Should take approximately 10
minutes. At the end of the semester and throughout the following years, you will be informed via e-mail
when to take a follow-up survey using the same ID and password that you created.

Remember, this survey is confidential. Your privacy will be protected to the maximum extent allowable by
law. Participation is voluntary, you may choose to not participate at all, or you may refuse to answer
certain questions or discontinue participation at any time. There are no known risks associated with
participation in this study. You will not benefit from participation in this study. However, your
participation in this study may contribute to the understanding of physical activity and other health related
behaviors of college students. A ticket number has been placed at the bottom of this page. This number
will allow you to enter the survey and create a userlD and password. These following steps will provide
you access to the survey.

1) First locate the 5-digit ticket number on the bottom of this page

2) Enter the website http://lsecommtechlab.msu.edu/

3) Under “New Users”, enter your five digit ticket number into the box on the webpage then click
“submit”.

4) Choose a userID and password for this survey then click “continue”.

PLEASE REMEMBER THE USERID AND PASSWORD. YOU WILL NEED THIS FOR
THE FOLLOW-UP SURVEY AT THE END OF THE SEMESTER.

5) Fill out the information page (i.e., What are the first two letters of your mother’s first name?) and
click “save”. This information will be used if you forget your userlD or password.

6) Complete the survey.

7) Close the webpage when the survey indicates you are done.

Upon completion of the survey, please retain the ticket number that you created. Take your ticket number
to Joann lanes in Room 3, 1M Circle in which you will be redeemed with a coupon for a the ice cream
cone at the Dairy Store.

If you have any questions about the study, please contact me (Joshua Ode) or James Pivamik (address: 3
1M Circle, East Lansing, MI 48824, phone: 353-3 520, e-mail: jimpiv@msu.edu). If you have questions or
concerns about your rights as a research participant, please feel free to contact Peter Vasilenko, Ph.D.,
Director of the Human Subject Protection Programs as Michigan State University: (517) 355-2180, fax:
(571)432-4503, e-mail: irb@msu.edu, or regular mail: 202 Olds Hall, East Lansing, MI 48824.

Thank you for your participation. Ticket Number
Joshua Ode
Doctoral Candidate: Department of Kinesiology

Phone: 355—4734
e-mail: odeioshu@msu.edu

123

Appendix B

Questionniare

1. What is your age?

17

18

19

20

21

22

23

24

25 or greater

r'e‘qv runes 9‘!”

2. What is your gender?
a. Male
b. Female

3. How do you describe yourself?

a. American Indian or Alaskan Native
b. Asian

c. Black or African American

d. Hispanic or Latino

e. Native Hawaiian or Pacific Islander
f. White

g. Other

4. What best describes your current living status

a. On-campus
b. Off-campus apartment/house with friends
c. Off-campus with parents/guardians
d. Off-campus alone
e. Other
5. Are you a Kinesiology Major?
a. Yes
b. No

6. During the past month, did you participate in any physical activities or exercises such as running,
calisthenics, golf, gardening, walking, etc.?
a. Yes
b. No

If yes, please continue to question #2
If no, please continue to question #13

7. Using the list provided, click on the physical activity or exercise you currently participate in most
often?

124

8.

10.

ll.

l2.

14.

15.

How many times per week do you usually participate in this activity
Less than 1

stress-99's»
\105Ut-hWN—t

Using 15 minute increments, how long do you usually participate in this activity?
15

30

45

60

75

90

greater than 90

(Present-rs»

Do you regularly participate in a second physical activity or exercise other than what was previously
mentioned?

a. Yes

b. No

If yes, please continue to question #6
If no, please continue to question #13

Using the list provided, click on the physical activity or exercise you currently participate in the
second most often?

How many times per week do you usually participate in this activity

a. Less than 1
b. l
c. 2
d. 3
e. 4
f. S
g. 6
h. 7

. Using 15 minute increments (15, 30, 45, 60, 75, etc.), how long do you usually participate in this

activity?

 

Do you regularly participate in a third physical activity or exercise other than what previously
mentioned?

a. Yes

b. No

If yes, please continue to question #10
If no, please continue to question #13

Using the list provided, click on the physical activity or exercise you currently participate in the third
most often?

125

16. How many times per week do you usually participate in this activity

17.

18.

19.

20.

21.

Less than 1

F'QPWPP-PP’P
mama-wro—

Using 15 minute increments (15, 30, 45, 60, 75, etc.), how long do you usually participate in this
activity?
a. 15
b 30
c. 45
d. 60
e 75
f. 90
g greater than 90

On average, how many hours do you watch TV per day?
Less than 1

1

2

3

4

5 or greater

”PP-.09"?

On average, how many hours of video games do you play per day?
Less than 1

1

2

3

4

5 or greater

finesse

On average, how many hours do you spend surfing the intemet, writing e-mails, or completing class
work on a computer per day (do not include playing video games)?

a. Less than 1
b. 1
c. 2
d. 3
e. 4
f. 5 or greater

Did you participate in any sponsored and/or club sports or activities (i.e., basketball, cheerleading,
marching band, etc.) during high school?

a. Yes

b. No

If yes, please continue to question #14
If no, go to question #

126

22. Using the list provided, click on all of the sponsored or club sports/activities that you participated in.
Baseball
Basketball
Cheerleading
Cross Country
Dance

Field Hockey
Football

Golf

Ice Hockey
Marching Band
Pom Pons
Rugby

Soccer

Softball
Swimming
Track
Volleyball
other

F-P'Peear—F‘r'r'a‘mrfieeop’e

16. How many times per week did you usually participate in these sports/activities?
Less than 1

rr'r'e‘qcrneensre

0

17. When you took part in these activities, how many minutes did you usually participate?
30

60

90

120

150

sees?!»

18. During high school, did you participate in any non-school sponsored leisure-time physical activities
or exercises such as running, sports, gardening, etc.?
a. Yes
b. No

19. Using the list provided, click on the leisure-time physical activities or exercises you participated in
most often during high school (choose no more than 3 activities).

127

20. On average, how many times per week did you participate in these activities combined?

a. 1

b. 2

c. 3

d. 4

e. 5

f. 6

g. 7

h. 8

i. 9

j. 10 or greater

21. When you took part in these activities combined, how many minutes did you usually participate?

a. 15

b. 30

c. 45

d. 60

e. 75

f. 90

g. greater than 90
24. What is your height in inches?
25. What is your weight in pounds?

26. What section of KIN 121 are you currently enrolled?

a. 001
b. 002
c. 003
d. 004
e. 005
f. 006

128

Appendix C

Log Transformation Data for Kcal/kg/week

Figure 8. Frequency distribution for the continuous physical activity outcome variable

 

 

 

 

 

 

 

 

 

 

 

 

ZOO-j
f—
150'1
>* _
0
C
G
3100
e 1
IL
h—r
50- _~
0 —l_ — — — l
0.00 200.00 400.00

kcalslkglwk

Table 15. Mean, skewness, and kurtosis data for the continuous physical activity
outcome variable

 

Mean 34.17
Standard Deviation 33.81
Skewness 3.29
Standard Error of Skewness 0.08
Kurtosis 20.50
Standard Error of Kurtosis 0.16

 

129

Figure 9. Frequency distribution for the log transformed continuous physical activity
outcome variable

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100-
80- ‘
fl1..
r—
p.
m
l— l—

6‘ so-
1:
or
3
cr
2
u.

40- ‘ _

F
— '1
20— 1 —
n i--1
0 I H I ’ ’ r I I
-1.00 0.00 1.00 2.00 3.00 4.00 5.00 6.00
Log of kca llkglweek

Table 16. Mean, skewness, and kurtosis data for the log transformed continuous physical
activity outcome variable

 

Mean 3.19
Standard Deviation 0.94
Skewness -0.62
Standard Error of Skewness 0.08
Kurtosis 0.76
Standard Error of Kurtosis 0.17

 

130

Appendix D
Results for Continuous Physical Activity Outcome Variable

Table 17. Descriptive data for the log kcal/kg/wk

 

 

Variable N mean (139 PA) SD (Log PA) fstatistic p-value
Gender
Male 314 3.51 0.87 60.562 0.000
Female 557 3.01 0.93
Class
Healthy Lifestyles 435 3.33 0.90 19.356 0.000
Communications 436 3.05 0.96
Major
Kinesiology 190 3.55 0.83 36.778 0.000
Other 677 3.09 0.95
Race
African American 56 2.90 1.26 3.210 0.040
Other 69 3.22 0.91
Caucasian 746 3.22 0.91
Age
22 or older 92 3.11 0.96 0.893 0.110
21 134 3.11 0.96
20 180 3.21 0.91
19 247 3.13 0.98
18 218 3.33 0.90
Living Status
On-campus 418 3.28 0.90 7.400 0.007
Off-campus 448 3.11 0.97
High school sports
4 or more 114 3.44 0.94 5.789 0.000
3 sports 218 3.30 0.90
2 sports 225 3.21 0.87
1 sport 177 3.07 0.90
0 sports 137 2.95 1.10
High school LTPA
vigorous 277 3.50 0.87 23.880 0.000
moderate 75 3.52 0.97
inactive 265 3.00 0.85
sedentary 253 2.96 0.99
TV
4 or more hours 66 3.07 1.03 2.190 0.088
3 hours 111 3.08 1.00
2 hours 227 3.13 0.99
less than 2 hours 467 3.27 0.88
VG
0 hours 615 3.09 0.97 13.800 0.000
1 hour or less 202 3.48 0.80
2 or more hours 54 3.26 0.91
BMI
Obese 44 3.24 0.92 0.918 0.400
Ovenrveight 171 3.27 0.98

Normal weight 637 3.16 0.93

131

Table 18. Multiple regression analysis results for the log kcal/kg/wk

 

 

R
Std. R square
Variable 8 Error P-value R square chagqe P-value
High school LTPA -0.166 0.025 0.000 0.256 0.066 0.066 0.000
Gender -0.463 0.062 0.000 0.333 0.111 0.048 0.000
Major 03% 0.072 0.000 0.383 0.147 0.036 0.000

High school sports -0.104 0.023 0.000 0.408 0.167 0.200 0.000
Television viewing 0.105 0.031 0.001 0.422 0.178 0.011 0.001

132

Appendix E

Results for Categorical Physical Activity Outcome Variable
Table 19: Descriptive data for meeting the recommendations for physical activity

 

 

Chi
Inactive Moderate Vigorous square
Variable N N % N % N % p-value
Gender
Male 321 48 15% 28 9% 245 76% 0.000
Female 587 143 25% 109 19% 335 57%
Class

Healthy Lifestyles 453 73 16% 58 13% 322 71 % 0.000
Communications 455 118 26% 79 17% 258 57%

Major
Kinesiology 193 19 10% 22 1 1% 152 79% 0.000
Other 71 1 172 24% 1 15 16% 424 60%

Race
African American 71 31 44% 14 20% 26 36% 0.000
Other 78 25 32% 10 13% 43 55%
Caucasian 759 135 18% 113 15% 511 67%

Age
22 or older 101 28 28% 18 18% 55 54% 0.256
21 142 37 26% 19 13% 86 61%
20 185 33 18% 26 14% 126 68%
19 255 54 21% 41 16% 160 63%
18 225 39 17% 33 15% 153 68%

Living Status
On-campus 437 81 18% 69 16% 287 66% 0.175
Off-campus 466 1 10 23% 68 15% 288 62%

Television viewing
4 or more hours 71 21 30% 11 15% 39 55% 0.067
3 hours 118 32 27% 19 16% 67 57%
2 hours 236 52 22% 40 17% 144 61%
less than 2 hours 483 86 18% 67 14% 330 68%
Video game use
0 hours 640 137 21 % 108 17% 395 62% 0.003
1 hour or less 208 36 17% 18 9% 154 74%
2 or more hours 60 18 30% 11 18% 31 52%
BMl

Obese 50 16 32% 7 14% 27 54% 0.376
Overweight 180 37 24% 24 13% 1 19 66%
Normal weight 657 134 20% 99 15% 424 65%
High school sports
4 or more 118 17 14% 13 11% 88 75% 0.002
3 sports 222 37 17% 30 14% 155 70%
2 sports 234 43 18% 37 16% 154 66%
1 sport 186 51 27% 30 16% 105 57%
0 sports 148 43 29% 27 18% 78 53%
High school LTPA
vigorous 281 36 13% 19 7% 224 80% 0.000
moderate 75 12 16% 8 1 1% 56 74%
inactive 276 58 21 % 62 23% 156 56%
sedentary 276 85 31 % 48 17% 143 52%

 

133

Table 20. Unadjusted odds ratios for meeting the recommendations for physical activity

 

 

OR OR
moderate vigorous
to p- to p-

Variable inactive 95 % Cl value inactive 95 % Cl value

Gender
Male 0.77 0.45 1.30 0.321 2.18 1.51 3.14 0.001
Female 1.00 1.00

Class
Healthy Lifestyles 1.19 0.76 1.86 0.453 2.02 1.44 2.82 0.001
Communications 1.00 1.00

Major
Kinesiology 1.73 0.90 3.34 0.102 3.25 1.95 5.40 0.001
Other 1.00 1.00

Race
African American 0.54 0.27 1.06 0.075 0.22 0.13 0.39 0.001
Other 0.48 0.22 1.04 0.062 0.45 0.27 0.77 0.003
Caucasian 1.00 1.00

Age
22 or older 0.76 0.36 1.61 0.474 0.50 0.28 0.89 0.018
21 0.61 0.29 1.25 0.175 0.59 0.35 1.00 0.049
20 0.93 0.47 1.86 0.840 0.97 0.58 1.64 0.919
19 0.90 0.48 1.66 0.730 0.76 0.47 1.21 0.240
18 1.00 1.00

Living Status
On-campus 1.38 0.89 2.14 0.154 1.35 0.97 1.88 0.073
Off-campus 1.00 1.00

Television viewing
4 or more hours 0.67 0.30 1.49 0.329 0.48 0.27 0.865 0.014
3 hours 0.76 0.40 1.46 0.414 0.55 0.34 0.885 0.014
2 hours 0.99 0.59 1.66 0.962 0.72 0.49 1.072 0.106
less than 2 hours 1.00 1.00

Video game use
0 hours 1.29 0.58 2.85 0.528 1.67 0.91 3.089 0.099
1 hour or less 0.82 0.32 2.09 0.676 2.48 1.25 4.927 0.009
2 or more hours 1.00 1.00

BMI
Obese 59% 0.23 1.49 0.267 0.53 0.28 1.02 0.057
Overweight 88% 0.49 1.56 0.658 1.02 0.67 1.54 0.939
Normal weight 1.00 1.00

High school sports
4 or more 1.22 0.51 2.90 0.656 2.85 1.51 5.41 0.001
3 sports 1.29 0.65 2.55 0.462 2.31 1.38 3.87 0.002
2 sports 1.37 0.71 2.63 0.343 1.97 1.19 3.26 0.008
1 sport 0.94 0.48 1.81 0.846 1.13 0.69 1.87 0.620
0 sports 1.00 1.00

High school LTPA
vigorous 0.93 0.48 1.81 0.841 3.70 2.38 5.76 0.000
moderate 1.18 0.45 3.09 0.735 2.77 1.41 5.47 0.003
inactive 1.89 1.14 3.13 0.013 1.60 1.07 2.39 0.023
sedentary 1.00 1.00

 

134

Table 21. Adjusted odds ratios for meeting the recommendations for physical activity

 

 

moderate vigorous
to p- to p-

Variable inactive 95 % Cl value inactive 95 % Cl value

Gender
Male 1.03 0.53 2.01 0.928 3.15 1.91 5.21 0.000
Female

Major
Kinesiology 1.70 0.85 3.40 0.900 3.08 1.78 5.33 0.000
Other

Race
African American 0.46 0.21 0.98 0.045 0.30 0.16 0.56 0.000
Other 0.50 0.22 1.16 0.107 0.55 0.30 1.02 0.057
Caucasian 1.00 1.00

Age
22 or older 0.93 0.41 2.14 0.872 0.63 0.32 1.22 0.172
21 0.68 0.32 1.45 0.318 0.74 0.42 1.31 0.305
20 1.05 0.51 2.14 0.900 0.92 0.52 1.62 0.769
19 0.98 0.52 1.85 0.944 0.72 0.43 1.19 0.198
18 1.00 1.00

High school sports
4 or more 1.09 0.44 2.75 0.848 2.50 1.22 5.11 0.012
3 sports 0.97 0.46 2.01 0.928 1.97 1.11 3.53 0.022
2 sports 1.17 0.58 2.35 0.664 2.04 1.16 3.58 0.013
1 sport 0.86 0.43 1.72 0.667 1.17 0.67 2.04 0.588
0 sports 1.00 1.00

High school LTPA
vigorous 0.79 0.40 1.58 0.505 2.90 1.79 4.69 0.000
moderate 0.86 0.31 2.39 0.769 2.44 1.18 5.06 0.017
inactive 1.57 0.92 2.66 0.096 1.43 0.92 2.23 0.109
sedentary 1 .00 1.00

Video game use
0 hours 1.16 0.45 2.99 0.765 3.02 1.38 6.61 0.006
1 hour or less 0.73 0.26 2.00 0.537 2.33 1.05 5.13 0.037

2 or more hours

135

Appendix F
Study Aim 3 Research Design
Appendix E] Research Design
In brief, the proposed research will begin with an undergraduate student
population at a large Midwestern University. Assessment of these students will include
an initial cross-sectional survey followed by a re-assessment prospectively at and interval
of 15 weeks followed by intervals of 6 months until the study participant is not enrolled
at the university. To achieve this projects specific aims, the analysis plan will include: 1)
analysis of cross-sectional (baseline) data, as well as 2) analysis of the longitudinal data.
Recruitment from a healthy lifestyles class student population and a communications
class student population made it possible to compare these students cross-sectionally,
after enrolhnent in their classes, and following enrollment in their classes. The first
cohort will include students enrolled in a healthy lifestyles class and communications
class recruited during the September 2007 and followed through December 2007.
Additional follow-up surveys will be disseminated every March and October thereafter.
The second cohort included students enrolled in these classes recruited in January 2008
and followed through April 2008. Additional follow-up surveys will be disseminated
every October and March thereafter. Additional cohorts will be added in subsequent
years. While these students will not be assigned at random to their classes, one of the
focal points of this research will be to assess the initial and long-term impact of
enrollment in a healthy lifestyles class (intervention group) compared to students enrolled
in a communications class (control group). This research protocol will have three major

elements: 1) completion of an anonymous online survey with standardized survey

136

questions and item sets and 2) a repetition of the online survey afier 15 weeks, 3)
repetition of the online survey at intervals of 6 months. Figures 10 and 11 represent the

timeline of this research protocol.

Figure 10: Timeline for data collection for study participants entering the cohort in
September

Baseline Follow-up 2 Follow-up 4 F ollow-up 6

 

Follow-up 1 F ollow-up 3 Follow-up 5

Figure l 1: Timeline for data collection for study participants entering the cohort in
January

Baseline Follow-up 2 F ollow-up 4 Follow-up 6

 

F ollow-up 1 F ollow-up 3 Follow-up 5

137

Appendix F.2 Study Population

The goal of this project is: l) to assess physical activity and health behaviors of
college students during four years of enrollment at a university 2) to improve regular
physical activity participation in college students with the expectation that this goal can
be achieved, in part, via classroom based educational interactions that occur during the
college years and 3) examine the long—term effects of a activity based class on physical
activity participation. To give this project focus and to ensure a timely completion,
undergraduate college students will be designated as the study population of interest. For
reason of feasibility, undergraduates enrolled in the healthy lifestyles class (KIN 121)
will be designated as the target population of primary interest because these students
were receiving educational interactions intended to influence their physical activity
levels. For a contrast with the experiences received during enrollment in the healthy
lifestyles class, a second target population will be designated and will consist of
undergraduate students enrolled in a communications class (COM 225). Students enrolled
in this class will not receive the activity and health component of the healthy lifestyles
class. An estimated 300 students will be enrolled in the healthy lifestyles class and 500

students enrolled in the communications class.

Appendix F .3 Eligibility criteria (inclusion/exclusion)
It was decided in advance that all enrolled students will be eligible for
participation in this research. That is, there will be no inclusion or exclusion criteria

applied to these undergraduate enrollees. Therefore, all 800 undergraduate students

138

enrolled in the healthy lifestyles or communications class will be eligible to participate in

this research.

Appendix F.4 Recruitment procedures

Due to the anonymity of the data collection, recruitment was completed according
to exempt approval status granted by the biomedical institutional review board at
Michigan State University. In brief, the steps of the recruitment process were as
followed.

9) Consent forms (Appendix A) will be distributed to all eligible study participants
enrolled in study during the first week of class.

10) The consent form will include a survey admission ticket and instructions needed
to access the on-line survey which will utilize the Longitudinal Surveillance
Engine (LSE).

11) Verbal instructions will be given by a study investigator to ensure understanding
of how to access the online survey.

12) Students assenting to participate will use their unique LSE admission tickets to
complete the online assessment.

13) Study participants will be given 10 days to complete the baseline questionnaire.
To encourage participation, a reminder e-mail will be sent to all study participants
on the 5th, 7‘", and 9th days of the 10 day period.

14) A second series of reminder e-mails will be sent during the 15th week of the study

period encouraging subjects to participate in the re-assessment survey.

139

15) Study participants will be given 10 days to complete the follow-up questionnaire.
A reminder e-mail will be sent to all study participants on the 5th, 7‘”, and 9th days
of the 10 day period.

16) Reminder e-mails will be sent every March and October encouraging subjects to
participate in the long-term re-assessment surveys.

17) Upon completion of the baseline survey and follow-up surveys, students will be
eligible to receive a coupon for free ice cream cone at the Dairy Store located at

Michigan State University.

Appendix F.5 Assessment Protocol:

This research project will utilize the Longitudinal Survey Engine (LSE) which was
recently developed at Michigan State University. The LSE is an intemet based survey
engine that offers anonymity and privacy for exploring topics that might be too sensitive
to effectively engage participants in scientifically meaningful disclosures. The LSE is
designed for large sample longitudinal, research protocols. Under the LSE protocols,
participants log in anonymously to create their own userID and passwords, and complete
an initial baseline survey. Thereafter, participants are invited to return for follow-up
surveys using the previously created userH) and passwords. The userID/password feature
allows the investigator to monitor individual and group patterns of health and behaviors
anonymously over time. The LSE can be menu-driven by authorized client-investigators
using major intemet browsers (Internet Explorer, Netscape, etc.). The main features of

the LSE used in the current investigation are described below:

140

g)

h)

i)

k)

1)

Each study participant will be provided with online access to the LSE
(http://lse.commtechlab.msu.edu/login.php)

The LSE software used a unique survey ID number to track each individual survey
protocol (e. g., an initial online baseline questionnaire or assessment followed by
longitudinal assessments), and also to track each participant's responses at baseline
and over time. (The LSE built a unique ID number to track both the survey protocol
and the individual participant. This ID number was called a 'coupon number.')

Prior to completion of the initial baseline assessment, the designated participant will
access the LSE website, key in his or her LSE coupon number, and will be prompted
to create his or her own userID and his or her own password. This will be followed
by prompting questions used to help the individual recall the userID and password if
these were lost or forgotten (e.g., 'What is the first initial of your mother's maiden
name (surname before marriage)?’ This approach will be used so designated
participants can log in anonymously without direct or indirect linkage between the
participant's identity and the survey responses.

Thereafier, the web browser will display a study disclosure statement inviting
participation. The participant will key the NEXT button to proceed to complete the
survey, or can log out at any time. The LSE was designed to allow the participant to
terminate participation and/or to skip over a survey item.

Survey elements used in this investigation will include true/false, multiple choice, and
short answer items.

The LSE will keep track of successive log-ins by designated participants, which

allows them to complete repeated measurements. Storage of the repeated measures

141

occurs within a database that encoded the userID and a date/time stamp. Number of
logins and the interval between repeated measures are controlled by the investigator
by Options given by the LSE. All responses are stored within a protocol-specific
EXCEL spreadsheet on the LSE secure server.

m) The LSE 'coupon management' system allows for the investigator to build an Excel
database from a unique study-participant ID generated for each study participant. The
database is imported to a word processing program for label-printing, which are

delivered to the designated participants via a sealed envelope to ensure anonymity.

Appendix F.7 Primary Outcome Variables

College leisure time physical activity participation: At entry into the cohort, all
study participants will be asked if s/he participated in any leisure-time physical activities
or exercises such as running, calisthenics, golf, gardening, walking, etc. during the past
month. If the study participant answers yes, a series of three remaining physical activity
questions will be asked to assess the type, fi'equency (sessions per week), and the
duration (average minutes per session) of activity. These questions are based on a similar
previously validated instruments 1’ 3’ 33. A metabolic equivalent value (MET) will be
assigned for each activity using the Compendium of physical activities to calculate
weekly caloric expenditure 2. If the study participant reports participating in a second or
third activity, the same series of questions will be repeated for each activity. The same
series of questions will be repeated at the end of the 15-week testing period and
throughout the follow-up surveys. These questions are similar to others used in physical

activity survey research 45‘ 50.

142

Appendix F.8 Exposure Variables

Gender: Gender will be defined as a dichotomous variable (male/female).

Class membership: Class membership will be defined as a dichotomous variable
(healthy lifestyles/commmlications).

Age: Age will be defined as a categorical variable (18 years, 19 years, 20 years,
21 years, 22 or more years)

Race: Race will be defined as a categorical variable (Caucasian, African
American, Other (e.g. Hispanic, Asian, Pacific Islander, Native American)).

Living Status: Current living status will be defined as a dichotomous variable
(living on—campus, living off-campus).

Major: Major will be defined as a dichotomous variable (kinesiology, other).

Class standing: Current class standing will be measured as a categorical variable
(freshman, sophomore, junior, senior).

Body Mass Index: Data on self-reported height and weight will be collected and
body mass index (kg/m2) will be calculated. Subjects with a BMI between 18.5 and 24.9
are considered normal weight, between 25 and 29.9 are overweight, and 30 or greater are
considered obese.

Sedentary Behaviors: Current sedentary behaviors will be assessed by
determining the number of hours per day spent: 1) watching television, 2) playing video
games, 3) using the computer for school related work, and 4) using the computer for non-
school related work.

High school sports: Participation in school sponsored sport activity will be

assessed by asking each subject if s/he participated in any school sponsored and/or club

143

sports or activities (e.g. basketball, cheerleading, marching band, etc.) during the last year
of high school. If the student participated in a school sponsored sport, s/he was first
asked to identify each high school sport in which s/he participated. This was followed by
asking: 1) how many months per year, 2) the average number of sessions per week, and
3) the average minutes per session s/he participated in these school-sponsored sports
combined. Using MET values for each physical activity, average weekly caloric
expenditure (kcal/kg/week) for a combination of all sponsored high school sports and
non-school sponsored physical activity during high school was calculated. In this study,
high school sport participation was defined as the number of high school sports played (0
sports, 1 sport, 2 sports, 3 sports, 4 or more sports).

High school leisure time physical activity participation (high school LTPA):
Participation high school leisure time physical activity will be assessed by asking if each
subject participated in any non-school sponsored leisure-time physical activities or
exercises (any activity that was not sponsored directly by the high school) such as
running, sports, gardening, etc. during high school during the last year of high school.
This was also followed by a series of questions asking: 1) how many months per year, 2)
the average number of sessions per week and 3) the average minutes per session s/he
participated in these non school-sponsored physical activities combined. Using MET
values for each physical activity, average weekly caloric expenditure (kcal/kg/week) for a
combination of all sponsored high school sports and non-school sponsored physical
activity during high school was calculated. Using the responses from this series of
questions, study participants were categorized as: l) sedentary (reported no non-school

sponsored sport activity, 2) inactive (reported some leisure time activity but did not meet

144

the recommendations for moderate or vigorous activity), 3) meeting the
recommendations for moderate physical activity, and 4) meeting the recommendations
vigorous physical activity. High school sponsored sport activity was not assessed the
same way as high school non-sponsored leisure time activity because the majority of
study participants who played a high school sport (90%) met the recommendations for
physical activity during high school sport participation.

Grade point average (GPA): Current Grade point average will be assessed as a
continuous variable.

Physical activity readiness: Physical activity readiness will be assessed as a
categorical variable (“I have never though about exercising”; “I have thought about
exercising a few times, but can’t get started”; “1 exercise every now and then.”; I try to
exercise regularly, but it’s hard to keep it up.”; “I am now exercising regularly.”)

Physical activity barriers: Barriers to physical activity will be assessed as a
categorical variable (School work; Lack of time; Weather; Motivation; Lack of facilities;
Enjoyment of exercise; Other; Barrier do not prevent me fro exercise).

Physical activity attitude: Attitude towards physical activity will be assessed on a
likert scale (l=strongly agree, 5=strongly disagree). The following statements will be
used to assess attitude towards physical activity: If I were to be physically active during
my free time on most days it would help me cope with stress; If I were to be physically
active during my free time on most days it would be fun.; If I were to be physically active
during my free time on most days it would help me meet friends; If I were to be

physically active during my free time on most days it would get or keep me in shape; If I

were to be physically active during my fi'ee time on most days it would make me feel

145

better about myself; If I were to be physically active during my free time on most days it
would give me more energy.; If I were to be physically active during my free time on
most days it would make me better in sports, dance, or other activities; If I were to be
physically active during my free time on most days it would make me hot and sweaty.

Physical activity self eflicacy: Physical activity self efficacy will be assessed on a
likert scale (l=strongly agree, 5=strongly disagree). The following statements will be
used to assess physical activity self efficacy: I can be physically active during my free
time on most days.; I can be physically active during my fiee time on most days even if I
could watch TV or play video games instead; I can be physically active during my free
time on most days even if it is very hot or cold outside; I can ask my best friend to be
physically active with me during my free time on most days.; I can be physically active
during my free time on most days even if I have to stay at home; I have the coordination
I need to be physically active during my free time on most days; I can be physically
active during my fi'ee time on most days not matter how busy my day is.

Physical activity behaviors: Physical activity behaviors will be assessed on a
likert scale (l=very easy, 5=very difficult; 1=strongly agree, 5=strongly disagree). The
following statements will be used to assess physical activity behaviors: For me to be
physically active during my free time on most days would be...; I have control over my
being physically active during my free time on most days.; I believe I have all the things I
need to be physically active during my free time on most days.; If I want to be I can be

physically active during my free time on most days.

146

10.

REFERENCES

Aadahl, M. and T. Jorgensen. Validation of a new self-report instrument for

measuring physical activity. Medicine and Science in Sports and Exercise.
35:1196-1202, 2003.

Ainsworth, B., W. Haskell, M. Whitt, M. Irwin, A. Swartz, S. Strath, W. O'Brien,
D. Bassett, K. Schmitz, P. Emplaincourt, D. Jacobs, and A. Leon. Compendium of

physical activities: an update of activity codes and MET intensities. Medicine and
Science in Sports and Exercise. 32:8498-8504, 2000.

Ainsworth, B., A. Leon, M. Richardson, D. Jacobs, and R. S. Paffenbarger.
Accuracy of the College Alumnus Physical Activity Questionnaire. Journal of
Clinical Epidemiology. 46: 1403-141 1 , 1993 .

ACSM's Guidelines for Exercise Testing and Prescription. 7th ed. Baltimore, MD:
Lippincott Williams & Williams, 2006.

American. American College Health Association. Healthy Campus 2010: Making
it Happen. Baltimore, MD: American College Health Association, 2002.

Anding, J. D., R. R. Suminski, and L. Boss. Dietary intake, body mass index,
exercise, and alcohol: Are college women following the dietary guidelines for
Americans? Journal of American College Health. 49:167-171, 2001 .

Booth, M., A. Okely, T. Chey, and A. Bauman. The reliability and validity of the
physical activity questions in the WHO health behaviour in schoolchildren
(HBSC) survey: a population study. British Journal of Sports Medicine. 35:263-
267, 2001.

Boreham, C., P. Robson, A. Gallagher, G. Cran, M. Savage, and L. Murray.
Tracking of physical activity, fitness, body composition, and diet from
adolescence to young adulthood: The Young Hearts Project, Northern Ireland.
International Journal of Behavioral Nutrition and Physical Activity. 1:14-21,
2004.

Boreham, C., J. Twisk, C. Neville, M. Savage, L. Murray, and A. Gallagher.
Associations between physical fitness and activity patterns during adolescence
and cardiovascular risk factors in young adulthood: the Northern Ireland Young
Hearts Project. International Journal of Sports Medicine. 23:822-826, 2002.

Bray, S. Self-efficacy for coping with barriers helps students stay physically

active during transition to their first year at a university. Research Quarterly for
Exercise and Sport. 78:61-70, 2007.

147

ll.

12.

l3.

14.

15.

16.

17.

l8.

19.

20.

21.

Bray, S. and H. Born. Transition to university and vigorous physical activity:
implications for health and psychological well-being. Journal of American
College Health. 52:181-188, 2004.

Brevard, P. B. and C. D. Ricketts. Residence of college students affects dietary
intake, physical activity, and serum lipid levels. Journal of the American Dietetic
Association. 96:25-28, 1996.

Burke, 8., A. Carron, and M. Eys. Physical activity context and university
student's propensity to meet the guidelines Centers for Disease Control and
Prevention/American College of Sports Medicine. Medical Science Monitor.
11:CR171-176, 2005.

Calfas, K., J. Sallis, and J. Nichols. Project Grad: two-year outcomes of a
randomized controlled physical activity intervention among young adults.
American Journal of Preventive Medicine. 18:28-36, 2000.

Caspersen, C., M. Pereira, and K. Curran. Changes in physical activity patterns in
the Unites States, by sex and cross-sectional age. Medicine and Science in Sports
and Exercise. 32:1601-1609, 2000.

Caspersen, C., K. Powell, and G. Christenson. Physical activity, exercise, and
physical fitness: definitions and distinctions for health-related research. Public
Health Reports. 100:126-131, 1985.

Craig, C., A. Marshall, M. Sjostrom, A. Bauman, M. Booth, B. Ainsworth, M.
Pratt, U. Ekelund, A. Yngve, J. F. Sallis, and O. Pekka. International Physical
Activity Questionnaire: 12-Country Reliability and Validity. Medicine and
Science in Sports and Exercise. 35:1381-1395, 2003.

Digest. Digest of Education Statistics, 2004. Available at:
http://nces.ed.gov/programs/digest/d04/ Accessed April, 2006.

Dinger, M. K. and A. Waigandt. Dietary intake and physical activity behaviors of
male and female college students. American Journal of Health Promotion.
11:360-362, 1997.

Douglas, K., J. L. Collins, C. Warren, L. Kann, R. Gold, S. Clayton, J. Ross, and
L. Kolbe. Results from the 1995 National College Health Risk Behavior Survey.
Journal of American College Health. 46:55-66, 1997.

Everhart, B., M. Kemodle, K. Ballard, C. McKey, B. Eason, and W. M. Physical

activity patterns of college students with and without high school physical
education. Perceptual and Motor Skills. 100:1 1 14-1 120, 2005.

148

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

Folsom, A., D. Jacobs, C. Caspersen, O. Gomez-Marin, and J. Knudsen. Test-
retest reliability of the Minnesota Leisure Time Physical Activity Questionnaire.
Journal of Chronic Disease. 39:505-51 1, 1986.

Ford, D. S. and C. R. Goode. African American college students' health behaviors
and perceptions of related health issues. Journal of American College Health.
42:206-210, 1994.

Gall, T., D. Evans, and S. Bellerose. Transition to first-year university: patterns of
change in adjustment across life domains and time. Journal of Social and Clinical
Psychology. 19:544-567, 2000.

Garman, J. F ., D. M. Hayduk, D. A. Crider, and M. M. Hodel. Occurrence of
exercise dependence in a college-aged population. Journal of American College
Health. 52:221-228, 2004.

Gyurcsik, N., S. Bray, and D. Brittain. Coping with barriers to vigorous physical
activity during transition to university. Family and Community Health. 27:130-
142, 2004.

Haase, A., A. Steptoe, J. F. Sallis, and J. Wardle. Leisure-tirne physical activity in
university students from 23 countries: associations with health beliefs, risk
awareness, and national economic development. Preventive Medicine. 39:182-

190, 2004.

Haberman, S. and D. Luffey. Weighing in college students’ diet and exercise
behaviors. Journal of American College Health. 46:189-191, 1998.

Harris, K. M., P. Gordon-Larsen, K. Chantala, and R. Udry. Longitudinal trends
in race/ethnic disparities in leading health indicators from adolescence to young
adulthood. Archives of Pediatrics and Adolescent Medicine. 160:74-81, 2006.

Henderson, 8., C. Haiman, L. Wilkens, L. Kolonel, P. Wan, and M. Pike.
Established risk factors account for most of the racial differences in
cardiovascular disease mortality. PLoS ONE. 2:e377, 2007.

Hensley, L. Current status of basic instruction programs in physical education at
American colleges and universities. Journal of Physical Education, Recreation &
Dance. 71:30-36, 2000.

Irwin, J. D. Prevalence of university students' sufficient physical activity: a
systematic review. Perceptual and Motor Skills. 98:927-943, 2004.

Jacobs, D., B. Ainsworth, T. Hartman, and A. Leon. A simultaneous evaluation of

10 commonly used physical activity questionnaires. Medicine and Science in
Sports and Exercise. 25:81-91, 1993.

149

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

Kahn, E., L. Ramsey, R. Brownson, G. Heath, E. Howze, K. Powell, E. Stone, M.
Rajab, and P. Corso. The effectiveness of interventions to increase physical

activity: A systematic review. American Journal of Preventive Medicine. 22:73-
107, 2002.

Kann, L., C. Warren, W. Harris, J. L. Collins, B. Williams, J. Ross, and L. Kolbe.
Youth Risk Behavior Surveillance - United States, 1995. MMWR CDC
Surveillance Summaries. 45: 1 -84, 1996.

Keating, X., J. Guan, J. Pinero, and D. Bridges. A meta-analysis of college
students' physical activity behaviors. Journal of American College Health.
54:116-125, 2005.

Kelley, G. A. and K. S. Kelley. Physical activity habits of African-American
college students. Research Quarterly for Exercise and Sport. 65:207-212, 1994.

Lamb, K. and D. Brodie. Leisure-time physical activity as an estimate of physical
fitness: a validation study. Journal of Clinical Epidemiology. 44:41-52, 1991 .

Landis, J. and G. Koch. The measurement of observer agreement for categorical
data. Biometrics. 33: 1 59-1 74, 1977.

Leighton, D. and H. Swerissen. Correlates of vigorous physical activity in young
adults during school transition. Australian Psychologist. 30:113-118, 1995.

Leslie, E., M. Fotheringham, N. Owen, and A. Bauman. Age-related differences
in physical activity levels of young adults. Medicine and Science in Sports and
Exercise. 33:255-258, 2001.

Leslie, E., M. Fotheringham, J. Veitch, and N. Owen. A university campus
physical activity promotion program. Health Promotion Journal of Austrailia. 10,
2000.

Leslie, E., N. Owen, J. Salmon, A. Bauman, J. F. Sallis, and S. K. Lo.

Insufficiently active Australian college students: Perceived personal, social, and
environmental influences. Preventive Medicine. 28:20-27, 1999.

Makrides, L., P. Veinot, J. Richard, E. McKee, and T. Gallivan. A cardiovascular
health needs assessment of university students living in residence. Canadian
Journal of Public Health. 89: 171-1 75, 1998.

Matthews, C., B. Ainsworth, C. Hanby, R. Pate, C. Addy, P. Freedson, D. Jones,

and C. Macera. Development and testing of a short physical activity recall
questionnaire. Medicine and Science in Sports and Exercise. 37:986-994, 2005.

150

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

Matton, L., M. Thomis, K. Wijndaele, N. Duvigneaud, G. Beunen, A. L.
Claessens, B. Vanreusel, R. Philippaerts, and J. Lefevre. Tracking of physical

fitness and physical activity from youth to adulthood in females. Medicine and
Science in Sports and Exercise. 38:] 114-1120, 2006.

McCullagh, P. Regression models for ordinal data. Journal of the Royal Statistical
Society. 42: 109-142, 1980.

MMWR. Youth risk behavior surveillance national college health risk behavior
survey — US, 1995. U. D. o. H. a. H. Services (Ed.), 1997, pp. 1-54.

Nelson, M., P. Gordon-Larsen, L. Adair, and B. Popkin. Adolescent physical
activity and sedentary behavior: patterning and long-term maintenance. American
Journal of Preventive Medicine. 28:259-266, 2005.

Paffenbarger, R. S. J., A. L. Wing, and R. T. Hyde. Physical activity as an index
of heart attack risk in college alumni. American Journal of Epidemiology.
108:161-175, 1978.

Page, R. Assessing college students' personal choices about health. College
Student Journal. 1:26-30, 1987.

Panel, B. Executive summary of the clinical guidelines on the identification,
evaluation, and treatment of overweight and obesity in adults. Archives of Internal
Medicine. 158:1855-1867, 1998.

Pearman, S., R. Valois, R. Sargent, R. Saunders, J. Wanzer Drane, and C. Macera.
The impact of a required college health and physical education course on the
health status of alumni. Journal of American College Health. 46:77-85, 1997.

Philippaerts, R., K. Westerterp, and J. Lefevre. Comparison of two physical
activity questionnaires with a tri-axial accelerometer to assess physical activity
patterns. International Journal of Sports Medicine. 22:34-39, 2001.

Philippaerts, R., K. Westerterp, and J. Lefevre. Doubly labeled water validation of

three physical activity questionnaires. International Journal of Sports Medicine.
20:284-289, 1999.

Pierce, E. F ., S. W. Butterworth, T. D. Lynn, J. O'Shea, and W. G. Hammer.
Fitness profiles and activity patterns of entering college students. Journal of
American College Health. 41 :59-62, 1992.

Pinto, B., N. Cherico, L. Szymanski, and B. Marcus. Longitudinal changes in
college students' exercise participation. Journal of American College Health.
47:23-27, 1998.

151

58.

59.

60.

61.

62.

63.

64.

65.

66.

67.

68.

69.

Pinto, B. and B. Marcus. A stages of change approach to understanding college
students’ physical activity. Journal of American College Health. 44:27-31, 1995.

Plasqui, G., A. Joosen, A. Kester, A. Goris, and K. Westerterp. Measuring free-
living energy expenditure and physical activity with triaxial accelerometry.
Obesity Research. 13:1363-1369, 2005.

Sallis, J. Age-related decline in physical activity: a synthesis of human and animal
studies. Medicine and Science in Sports and Exercise. 32: 1598-1600, 2000.

Sallis, J ., K. Calfas, J. Nichols, J. Sarkin, M. F. Johnson, S. Caparosa, S.
Thompson, and J. Alcaraz. Evaluation of a university course to promote physical
activity: project GRAD. Research Quarterly for Exercise and Sport. 70: 1-10,
1999.

Schoeller, D. and E. van Santen. Measurement of energy expenditure in humans
by doubly labeled water method. Journal of Applied Physiology. 53:955-959,
1982.

Schuit, A., E. Schouten, K. Westerterp, and H. Saris. Validity of the Physical
Activity Scale for the Elderly (PASE): According to energy expenditure assessed

by the doubly labeled water technique. Journal of Clinical Epidemiology. 50:541-
546, 1997.

Shephard, R. Limits to the measurement of habitual physical activity by
questionnaires. British Journal of Sports Medicine. 37: 197-206, 2003.

Sparling, P. B. and T. K. Snow. Physical activity patterns in recent college
alumni. Research Quarterly for Exercise and Sport. 72:200-205, 2002.

Stoops, N. Educational Attainment in the United States: 2003 . Washington DC:
US Census Bureau, 2004.

Suminski, R. R., R. Petosa, A. C. Utter, and J. J. Zhang. Physical activity among
ethnically diverse college students. Journal of American College Health. 51:75-
80, 2002.

Telama, R. and X. Yang. Decline of physical activity from youth to young
adulthood in Finland. Medicine and Science in Sports and Exercise. 32:1617-
1622, 2000.

Telarna, R., X. Yang, J. Viikari, I. Valimaki, O. Wanne, and O. Raitakari.

Physical activity from childhood to adulthood. American Journal of Preventive
Medicine. 28:267-273, 2005.

152

70.

71.

72.

73.

United. United States Department of Health and Human Services. Physical
activity and health: a report of the Surgeon General, 1996

United. US Department of Health and Human Services. Youth risk behavior
surveillance - US, 2001. MMWR. 51 :1-64, 2002.

United. US Department of Health and Human Services. Prevalence of physical
activity, including lifestyle activities among adults - US, 2000-2001. MMWR.
52:764-769, 2002.

van Mechelen, W., J. Twisk, G. Post, J. Snel, and H. C. Kemper. Physical activity

of young people: the Amsterdam Longitudinal Growth and Health Study.
Medicine and Science in Sports and Exercise. 32:1610-1616, 2000.

153